Computer vision · 7 min read
Five computer vision applications in production, on the cameras a site already owns
Detect, count, inspect, track and read are the five jobs a fixed camera can be given, and each is a different question with a different label behind it.
Ayman Quadir · Oct 4, 2026
Computer vision · 6 min read
Computer vision projects worth building on the cameras you already have
A plant, a utility, a grower, a store and a warehouse each have a project that starts on an existing camera, produces a decision, and has someone to act on it.
Ayman Quadir · Oct 4, 2026
Computer vision · 6 min read
How to choose an object detection model architecture for a camera on your own site
Start from the decision the plant has to make, then the box beside the recorder the model has to fit, and only then the family. The benchmark comes last.
Andreas Ohrvall · Oct 4, 2026
Computer vision · 6 min read
Image classification workflows, when a tag on the frame is enough and a box is too much
One panel per frame at the inspection station wants a tag rather than a box, and a tag costs a fraction of what boxes cost to label and to check.
Sheikh Srijon · Oct 4, 2026
Computer vision · 6 min read
Improving a custom object detection model, the tactics that still move the number
Two thousand frames from the real line beat ten thousand from a bench, a written box rule beats a talented labeler, and curation is a job that never finishes.
Rajiya Sultana · Oct 4, 2026
Computer vision · 6 min read
Object detection in production, what a detector returns and what it learns from
For every frame off the yard camera the detector returns a list of boxes with a class and a score, and the score is a ranking rather than a probability.
Esdras Ntuyenabo · Oct 4, 2026
Computer vision · 6 min read
How to evaluate object detection models fairly, and why two mAP numbers are rarely comparable
A vendor's score and yours are answers to different questions. The test set, the calculation and the inference conditions each move the number on their own.
Rob Hickey · Oct 4, 2026
Computer vision · 6 min read
Object detection vs image classification vs keypoint detection, and how to choose before you label
Is the frame defective, where are the defects, where are the joints of the arm. Three questions on one cell camera, and the answer fixes the label format.
Sheikh Srijon · Oct 4, 2026
Computer vision · 6 min read
Overfitting in computer vision, or how a model learns the warehouse instead of the pallet
A pallet detector that shines at its own warehouse and fails at the next one learned the lighting, and the training versus validation gap is the tell.
Rob Hickey · Oct 4, 2026
Computer vision · 6 min read
What pose estimation gives a site camera that a box cannot
A box says a worker is there. Keypoints say the worker is bending, reaching, or has a wrist a metre from a live terminal, and that is the safety question.
Stephen Biswas · Oct 4, 2026
Computer vision · 6 min read
How to train, evaluate and deploy a computer vision model on a camera you already own
From the first labeled frame off a dock camera to a model watching that dock, in the order the work really happens, with the retrain built in from day one.
Rob Hickey · Oct 4, 2026
Computer vision · 6 min read
Transfer learning starting points, why a small nearby dataset beats a huge general one
Four checkpoints for the same hard hat model, and the small one trained on people won. Pick a base model by what it has seen, then feed it the site's frames.
Esdras Ntuyenabo · Oct 4, 2026
Computer vision · 6 min read
What a convolutional neural network does with a frame from a line camera
One small filter slides across every position in the frame, and that single design decision is why the network fits on a box beside the recorder.
Finn Ellingwood · Oct 4, 2026
Computer vision · 7 min read
YOLO training best practices, what decides whether the model survives the factory floor
Frames from the floor itself, a labeling spec written before the first box, per-class recall over the average, and a loop that keeps running after go-live.
Stephen Biswas · Oct 4, 2026
Labeling · 7 min read
Aerial dataset augmentation for drone frames where there is no up
A tower seen straight down has no top or bottom, so rotations and both flips are safe. Scale for altitude, brightness for sun, move every box with its pixels.
Esdras Ntuyenabo · Oct 3, 2026
Computer vision · 6 min read
What CLIP is and why typing a description finds the frame
A shared space for pictures and words lets an operator type a sentence and get the matching frames from a recorder. A way to look, and the boxes come after.
Rob Hickey · Oct 3, 2026
Labeling · 6 min read
A collaborative data annotation workflow run as a pipeline
Batch the bottling line's frames by camera and shift, assign so nothing is boxed twice, attach the guideline, review every label, train on the approved set.
Rajiya Sultana · Oct 3, 2026
Labeling · 7 min read
Dataset health check for computer vision, what to look at before anything trains
A scratch dataset where every scratch sits in the centre of the frame will train a model that looks in the centre. Five counts to read before the first epoch.
Rajiya Sultana · Oct 3, 2026
Labeling · 6 min read
A dataset quality audit for object detection when mAP hides a weak class
The PPE camera at the rig site scored well on average and missed most bare heads. Audit the labels per class, fix the boxes in review, and retrain on the fixes.
Rob Hickey · Oct 3, 2026
Computer vision · 6 min read
An end-to-end object detection workflow starts with decisions, and the first frame is labeled last
Tag, box, mask or keypoints. A class list with occlusion and size rules. A metric and an acceptance sentence. Then, and only then, the first camera's frames.
Ayman Quadir · Oct 3, 2026
Labeling · 7 min read
Few-shot image labeling on a vial tray, one box drawn and the rest proposed
Draw one box on a vial, let the tray fill with proposals, and give a person the last word on each before the batch commits. The fastener bin is the hard case.
Rajiya Sultana · Oct 3, 2026
Labeling · 7 min read
How to evaluate an image annotation partner before sending them a year of footage
Six questions for a labeling partner, from how accuracy is measured to who settles a disagreement, asked before the inspection footage leaves the building.
Ayman Quadir · Oct 3, 2026
Labeling · 7 min read
Brightness and contrast adjustment for annotation, labeling the defect the camera barely shows
A faintly hot bushing on a night thermal frame is there and the labeler cannot see it. Turn the display up, draw the box, and leave the training pixels alone.
Stephen Biswas · Oct 3, 2026
Labeling · 6 min read
Image dataset search and filtering to find the frames a model never saw
The yard dataset has hundreds of daytime forklifts and almost none after dark. Filter by class and count, ask for frames like a dock at night, label the gap.
Ayman Quadir · Oct 3, 2026
Computer vision · 7 min read
Image recognition AI, explained on the cameras a site already owns
A tag, a box, a mask or keypoints, on a substation camera, a line camera and an aisle camera. How a model gets there, and what its score does not promise.
Esdras Ntuyenabo · Oct 3, 2026
Computer vision · 7 min read
Image segmentation and the questions on a weld bead that only a mask can answer
A box finds the weld. A mask measures it. Semantic, instance and panoptic told apart on a weld cell and a board camera, with what each costs to label.
Esdras Ntuyenabo · Oct 3, 2026
Computer vision · 7 min read
Monocular vs stereo depth estimation for a robot that has to stop in time
A stop rule needs a distance in metres. Monocular depth gives a ranking, stereo gives metres and loses them on a blank wall. The rule decides which one.
Rob Hickey · Oct 3, 2026
Computer vision · 7 min read
Object counting with computer vision, three kinds of count and why each needs its own pipeline
Forty boxes on a pallet, unique trucks through a gate over a shift, people in a checkout zone right now. Same word, three pipelines, and two ways to be wrong.
Finn Ellingwood · Oct 3, 2026
Labeling · 7 min read
Occlusion in computer vision, training for the half the camera can see
A forklift behind a rack, a shopper behind a display, a part under the robot arm. On a fixed camera the partial view is the normal one, and the labels say so.
Stephen Biswas · Oct 3, 2026
Computer vision · 7 min read
What optical character recognition is and where OCR breaks on a real camera
A packaging line camera is not a scanner. Acquisition, detection, recognition and validation on a lot-code label, and the doubtful read goes to a person.
Finn Ellingwood · Oct 3, 2026
Computer vision · 6 min read
Oriented bounding box detection, when the box needs to turn with the object
A straight box on a panel tilted at forty-five degrees is mostly background, and on a packed array the boxes overlap. A rotated box fixes that, at a price.
Stephen Biswas · Oct 3, 2026
Labeling · 6 min read
Passive image collection at the edge, letting the camera build its own training set
Leave the dock camera sampling over a weekend, drop near-duplicates at the box, pull the rare scenes with a typed description, and send the rest to review.
Andreas Ohrvall · Oct 3, 2026
Labeling · 6 min read
Polygon annotations for object detection when the model only draws boxes
A part lying diagonal on the bench gets a box that is mostly bench. Rotate a polygon and the box stays tight, which is why the extra labeling minute pays back.
Finn Ellingwood · Oct 3, 2026
Computer vision · 7 min read
Pose estimation algorithms and what keypoints give a safety camera that boxes do not
A box says a person is beside the press. Keypoints say where the wrist is. How joints are learned as heatmaps, and which questions are worth the cost of asking.
Rob Hickey · Oct 3, 2026
Labeling · 7 min read
How to reduce dataset size without losing accuracy on recorder footage
A dataset built from a recorder is mostly the same frame repeated. Which frames can go, which must stay, and how to know the model did not notice.
Sheikh Srijon · Oct 3, 2026
Labeling · 7 min read
Representative training data for computer vision on a site camera
A PPE and fall model that scored well at the pilot site missed the crew at the second. The labeled set had one kind of worker in it. The site had every kind.
Ayman Quadir · Oct 3, 2026
Labeling · 7 min read
Stop sign detection edge cases and the long tail of a simple object
A stop sign is the easiest class there is until it is snowed on, held by a crossing guard or printed on a bus. The tail is covered one doubted frame at a time.
Stephen Biswas · Oct 3, 2026
Computer vision · 7 min read
Traditional computer vision vs deep learning, when a threshold is enough and when the line needs a model
A colour threshold checks caps until the lamp changes. A scratch on dark paint has no rule anyone can write. Most lines end up running both, in that order.
Andreas Ohrvall · Oct 3, 2026
Labeling · 7 min read
Splitting camera footage into train, validation and test so the score means something
Frames a second apart from one camera land on both sides of a random split and the test score lies. Split by camera and by day, and hold out the newest camera.
Sheikh Srijon · Oct 3, 2026
Computer vision · 6 min read
Transfer learning from a pretrained checkpoint, why a few hundred frames are enough when you start warm
A pretrained backbone already knows edges and shapes. A few hundred reviewed frames from a site camera teach the head what a guardrail is. Retrains begin warm.
Rob Hickey · Oct 3, 2026
Labeling · 6 min read
Checking a public dataset's provenance before it trains your model
Who made it, what the licence allows, what a sample of its boxes shows, and how far its frames sit from your gate camera. Four checks before a download trains.
Sheikh Srijon · Oct 3, 2026
Computer vision · 7 min read
What a vision transformer does with a frame, and what changes in monitoring when you use one
The frame becomes patches, every patch looks at every other, and a tower over a hedge gets easier. The cost grows with resolution. The review queue stays.
Rob Hickey · Oct 3, 2026
Labeling · 6 min read
Visualizing dataset embeddings for quality control before training
A map of the depot dataset shows a pallet hiding in the forklift cluster, a knot of frames from one quiet minute, and a blank where the night shift should be.
Sheikh Srijon · Oct 3, 2026
Labeling · 6 min read
Zero-shot detection vs custom training, where a prompted model runs out
A prompted model boxes the forklift first try and has never seen a wafer scratch. Test on a small labeled set, then let corrections train the model that has.
Andreas Ohrvall · Oct 3, 2026
Labeling · 7 min read
AI data labeling workflows, three ways to label footage and when each one pays
A pipeline right-of-way survey labeled three ways: every box by hand, Lexi proposing and a person checking, and synthetic frames for the leak nobody has filmed.
Rajiya Sultana · Oct 2, 2026
Labeling · 6 min read
Annotation analytics, the numbers a labeling queue produces besides labels
Three labelers on a month of warehouse footage produce boxes, and also a throughput, an acceptance rate and a map of where the rejections cluster.
Rajiya Sultana · Oct 2, 2026
Labeling · 7 min read
Annotation format conversion between COCO, YOLO and CVAT without losing a box
Three years of line inspection labels from two tools arrive in three formats. The boxes that shift are the ones nobody draws on a frame before training.
Esdras Ntuyenabo · Oct 2, 2026
Operations · 6 min read
Batch video analysis of archived drone survey footage without a notebook open
Three seasons of right-of-way flights in a cloud bucket. Import the originals, sample the frames, run the model, and review only what it doubted.
Stephen Biswas · Oct 2, 2026
Labeling · 6 min read
Blur augmentation in computer vision, training for the blur the camera will produce
A camera on a vibrating gantry, an autofocus that hunts, fog on the housing. Train on the blur the line makes, never on the class it would erase.
Rob Hickey · Oct 2, 2026
Operations · 6 min read
Semantic search across a hundred live camera feeds with one sentence
An operator types what the rare scene looks like and the yard cameras return the frames that match. Search finds candidates; a person confirms them.
Sheikh Srijon · Oct 2, 2026
Operations · 7 min read
Computer vision event logging that keeps the frame with the prediction
A missed bone fragment on the night shift can only be explained if the frame, the prediction, the model version and the lot were logged together.
Rob Hickey · Oct 2, 2026
Labeling · 7 min read
Data annotation for computer vision is the problem statement, written in boxes
On a warehouse frame, the pallet nobody boxed teaches absence and the loose box teaches where a pallet ends. The schema is the spec, and QA is what holds it.
Sheikh Srijon · Oct 2, 2026
Labeling · 7 min read
Data augmentation for object detection on a fixed inspection camera
A stamping line camera varies in known ways. Conveyor speed changes scale and blur, a relamp changes the light. Augment for those and move the boxes too.
Stephen Biswas · Oct 2, 2026
Labeling · 7 min read
Handling imbalanced classes when the defect is one part in a thousand
A pinched cable once in a thousand housings. Why the loss can afford to ignore it, how to collect the rare frames on purpose, and the one honest metric.
Rob Hickey · Oct 2, 2026
Labeling · 6 min read
Hard hat detection datasets, helmet or head is the label that matters
A rig site camera labeled helmet, head and person. The head class is the one that makes a compliance rule possible, and the one most datasets leave out.
Ayman Quadir · Oct 2, 2026
Labeling · 6 min read
How to identify mislabeled images before the next retrain learns them
A shelf dataset had bleach boxed as beverages for a month and validation never noticed. Find the frames where the model disagrees with the label.
Sheikh Srijon · Oct 2, 2026
Labeling · 6 min read
Image annotation tools, box, polygon or mask is decided by the question
A cracked insulator, a corroded patch and a person in a restricted zone each force a different geometry. The wrong shape means relabeling everything.
Rob Hickey · Oct 2, 2026
Labeling · 6 min read
Image preprocessing vs data augmentation, and why only one runs at the camera
Resize, orientation and normalisation have to match between the labeled frames and the runner. Augmentation runs in training only, never at the camera.
Stephen Biswas · Oct 2, 2026
Labeling · 7 min read
Image resizing for computer vision, and the resize that squashes the forklift
A wide yard frame goes into a square input. Pad rather than stretch, move the boxes with the pixels, and treat a changed interpolation on the runner as a step.
Esdras Ntuyenabo · Oct 2, 2026
Labeling · 6 min read
Instance segmentation data labeling where two parts touch
Two fillets overlap on a blue conveyor and a corroded patch fades into clean steel. The label is a decision about where one thing ends, made before the drawing.
Finn Ellingwood · Oct 2, 2026
Operations · 6 min read
How big a computer vision model the line camera actually needs
Nano to large on one weld cell camera. What a bigger model buys on look-alike defects, what it costs on the runner, and choosing by the device it runs on.
Rob Hickey · Oct 2, 2026
Edge · 7 min read
What the ONNX file is, and why it sometimes loads and returns nothing
One exported file runs on the Jetson at the line, the Pi at the second site and the rack server. When it loads and draws no boxes, check the operator set first.
Finn Ellingwood · Oct 2, 2026
Labeling · 7 min read
Outsourced data labeling for computer vision, what to write down first
Before anyone else boxes your packing station frames, the guideline, the gold set and the checks on each delivery are what decide whether the labels are usable.
Sheikh Srijon · Oct 2, 2026
Operations · 7 min read
Drawing polygon regions of interest for the zone a rule applies to
The press zone, the dock lane and the checkout queue are polygons on a frame. Perspective makes them trapezoids, and a nudged camera moves them all at once.
Esdras Ntuyenabo · Oct 2, 2026
Operations · 6 min read
Role-based access control for computer vision labels and cameras
A contractor who labels one project, a reviewer who approves but cannot edit the class list, an admin who deploys, and why deny is the default for footage.
Ayman Quadir · Oct 2, 2026
Labeling · 7 min read
SAM assisted segmentation labeling, click the patch and correct the mask
A labeler clicks a corroded patch, a mask appears, and the job becomes correcting an edge instead of tracing one. The person earns it where the proposal fails.
Stephen Biswas · Oct 2, 2026
Operations · 7 min read
Storing a year of computer vision predictions without drowning in them
A battery line's tab welder produces a box on every frame all year. Keep the flagged events with the lot number and the frame, and let the rest go.
Rajiya Sultana · Oct 2, 2026
Operations · 7 min read
Streaming computer vision predictions into the event bus a plant already runs
The capper camera and the labeler camera on a filling line publish each detection as a timestamped event by webhook, and the plant's own bus carries it.
Andreas Ohrvall · Oct 2, 2026
Labeling · 7 min read
Synthetic image data for the defect that happens once in a million units
A crushed carton corner too rare to photograph. Rendered variants across lighting and severity, mixed with real frames, evaluated only on real ones.
Sheikh Srijon · Oct 2, 2026
Labeling · 7 min read
Uploading a week of images, videos and annotations without losing the rare frames
A week of the scanner camera is mostly the same carton. Import the originals, sample the routine, keep the jam at 3 am whole, bring old labels across intact.
Finn Ellingwood · Oct 2, 2026
Operations · 6 min read
Ask the footage a question with a VLM, but let the trained model count
What does the gauge on the pump skid read, what does the label at the dock say. A question gets an answer with frames behind it. Counting all shift is a rule.
Ayman Quadir · Oct 2, 2026
Operations · 6 min read
Comparing two model versions visually on the same shelf frame before rollout
Version 4 of the shelf gap model finds the shadowed gap version 3 missed, and invents one on the dark packs. Overlay both on one frame and look.
Rajiya Sultana · Oct 2, 2026
Operations · 7 min read
AGPL-3.0 licensing risk for computer vision teams serving a model
A camera streaming to a served detector is the network interaction the licence was written for. What a legal review will ask, and why to pick the weights first.
Ayman Quadir · Oct 1, 2026
Operations · 7 min read
Cloud vs owned GPU inference for computer vision, worked out per camera hour
A plant on three shifts and a retailer with cameras spread across stores get different answers from one sum, and footage leaving the building is a cost too.
Ayman Quadir · Oct 1, 2026
Operations · 7 min read
Computer vision heatmaps drawn from the aisle cameras a store already has
Footpoints from every tracked box, aggregated over a day and mapped onto the floor plan, show where footfall goes. A camera nudged in cleaning shifts the map.
Rajiya Sultana · Oct 1, 2026
Operations · 7 min read
Computer vision in data analytics, the camera as a table analysts can join
Aisle cameras become rows with timestamps: counts, dwell times, zone events. Join them to the till and a promotion shows in the aisle before the sales.
Sheikh Srijon · Oct 1, 2026
Edge · 6 min read
Computer vision on multiple video streams from one runner
Twenty cameras, one box beside the recorder. Fair sampling, a slow camera that drops its own frames, an unplugged one that stalls nobody, one heartbeat each.
Andreas Ohrvall · Oct 1, 2026
Operations · 6 min read
Computer vision safety triggers that stop the machine when what should be there is not
Hands off the palm buttons, a spotter missing, a person down on the belt. The trigger is an absence held past a window, and a dropped frame never stops a line.
Rajiya Sultana · Oct 1, 2026
Operations · 6 min read
Vision AI team structure, and who owns the model after launch
The pilot worked and the ML team moved on. The second year belongs to a builder, an operator who reviews and approves, and someone who makes site two a repeat.
Ayman Quadir · Oct 1, 2026
Operations · 6 min read
How to convert DAV footage to MP4 for a dataset without re-encoding
A warehouse recorder exports DAV that nothing opens. Remux the stream to MP4 untouched, batch the folder, keep the timestamps, and import it once.
Esdras Ntuyenabo · Oct 1, 2026
Edge · 7 min read
CPU vs GPU for computer vision inference, and when a CPU is enough
A cap station checked every couple of seconds and a shelf camera sampled every few minutes both run on a CPU. The GPU earns its keep when the load piles up.
Andreas Ohrvall · Oct 1, 2026
Edge · 6 min read
Deploying computer vision models to edge devices beside the camera
A robot cell that cannot wait for a round trip and an orchard with no uplink. Export by device, run beside the recorder, and get the next version out there.
Andreas Ohrvall · Oct 1, 2026
Edge · 6 min read
Edge computer vision for industrial automation, where the camera is the heaviest sensor on the plant network
A plant network built for PLC tags was never built for video. The runner beside the recorder turns frames into events, and only events cross the network.
Andreas Ohrvall · Oct 1, 2026
Operations · 7 min read
Embedding based anomaly detection for the frame the model has never seen
A car park camera meets its first snowplough and a line camera meets the maintenance crew. Distance from the usual frames is one check that finds them.
Rob Hickey · Oct 1, 2026
Operations · 6 min read
What end-to-end computer vision has to mean for a camera fleet
Four teams, four handoffs, and a correction that never reaches the next model. End to end is a test you can run, and most workflows fail it at the seam.
Ayman Quadir · Oct 1, 2026
Operations · 6 min read
How to choose a camera for computer vision before you collect a frame
Work back from the smallest thing the model must see, then shutter, lamp and interface. A camera swapped later is a sensor replacement with a relabel.
Stephen Biswas · Oct 1, 2026
Operations · 6 min read
How to count objects in a zone, from a checkout queue polygon to a staging bay
Draw the polygon once, keep the right class, test each box's bottom centre, alert on a count held for a window. Then watch the camera; the polygon will not.
Finn Ellingwood · Oct 1, 2026
Operations · 6 min read
How to increase inference speed for computer vision without a bigger GPU
A defect model that cannot keep pace with the press gets fixed by resolution, sampling and change gating long before the hardware budget is touched.
Andreas Ohrvall · Oct 1, 2026
Operations · 6 min read
Industrial machine vision lighting and optical filters decide what the camera sees before any model does
The scratch an inspector finds by tilting the part under a window is invisible under the wrong lamp. Light it once, and treat a lamp change as a new camera.
Stephen Biswas · Oct 1, 2026
Edge · 6 min read
Inference latency in computer vision, and the budget between the camera and the reject arm
A carton takes a fixed time to reach the reject arm from the lens. Everything from capture to the signal has to fit inside it, and most of it is not the model.
Esdras Ntuyenabo · Oct 1, 2026
Operations · 7 min read
Migrating computer vision datasets between platforms without losing a box
Export as COCO, YOLO or CVAT XML, check where each format puts the corner and the class, import with nothing re-encoded, compare boxes before training.
Sheikh Srijon · Oct 1, 2026
Operations · 6 min read
How to monitor inference health when nothing throws an error
A camera that stops sending, a runner that falls behind, a correction rate climbing on one site. Three failures with no error message and three different fixes.
Rob Hickey · Oct 1, 2026
Operations · 7 min read
Motion detection in computer vision that ignores the parked truck
A dock camera that alerts on the truck that just arrived and stays silent on the one parked since 5 am, with wind in the trees trained out.
Rob Hickey · Oct 1, 2026
Operations · 6 min read
Object measurement with computer vision, turning pixels into millimetres without a calibrated camera
A tile of known size on the belt gives each frame its scale. A rotated box follows the fillet, a polygon follows its edge, and a nudged camera breaks the plane.
Sheikh Srijon · Oct 1, 2026
Operations · 7 min read
OCR on video, reading a container number from a moving truck
A yard camera tracks each container as it passes, reads the ID panel on the sampled frames, and takes a vote. One blurry frame never becomes a wrong record.
Rajiya Sultana · Oct 1, 2026
Operations · 7 min read
QR code detection with computer vision when the code on the bin will not read
Finder patterns and the quiet zone for whoever is debugging a failed read, a detector that crops the code from a wide frame, and the result logged per tote.
Finn Ellingwood · Oct 1, 2026
Operations · 6 min read
How to reduce class flickering in object detection on video
A shelf facing that reads empty, stocked, empty every few frames is a per-frame model at a boundary. Vote over the track, then review the frames that flickered.
Rob Hickey · Oct 1, 2026
Edge · 6 min read
RTSP stream processing for computer vision, and the three ways a camera feed fails without an error
A feed that falls behind, a reboot that ends the frames with no error, a stream with the wrong clock. Sample every two seconds and give each camera a pulse.
Esdras Ntuyenabo · Oct 1, 2026
Operations · 7 min read
Thermal infrared imaging for computer vision on a transformer yard
A thermal camera sees the hot bushing at 2 am and the person at the fence in smoke. It needs its own labels and a recalibration when the sensor is swapped.
Sheikh Srijon · Oct 1, 2026
Operations · 6 min read
When to add an LLM to a vision pipeline and ask the footage a question
A detector boxes the crushed carton on the depot conveyor. Whether the label is still readable, and where the parcel goes now, is a question for the footage.
Ayman Quadir · Oct 1, 2026
Operations · 6 min read
YOLO versus vision language models, decided by the failure you can afford
A detector gives the same box for the same frame, every two seconds, forever. A language model answers a question nobody wrote a class for. The dock needs both.
Rob Hickey · Oct 1, 2026
Industries · 6 min read
Perimeter security with fixed cameras, object detection and a drone sent to look
A frame every two seconds is enough to catch a person at the fence, a CPU is enough to run it, and the drone is the second look rather than the detector.
Andreas Ohrvall · Sep 30, 2026
Operations · 6 min read
Camera calibration for computer vision, and why the part at the edge of the frame measures wrong
A straight edge bows at the corner of the frame, so a part that passes in the centre fails at the edge. Calibrate once, and again the day the lens changes.
Stephen Biswas · Sep 30, 2026
Industries · 7 min read
Food service QA with a camera over the tray packing line
Every component on the tray gets a box, the missing one is flagged before the sealer, and the alert count is read against the line's own history.
Ayman Quadir · Sep 30, 2026
Operations · 6 min read
Evaluating a computer vision platform once the pilot ends
The feature spreadsheet cannot tell you which platform survives the second year. Walk the lifecycle instead, from import to a retrain that keeps the old frames.
Ayman Quadir · Sep 30, 2026
Industries · 7 min read
Railway safety with trackside cameras, zones and a signaller who can live with the alerts
People and vehicles boxed, the track bed and crossing drawn as zones, the frame sent to the control room, and a false alarm rate a signaller will keep reading.
Rajiya Sultana · Sep 30, 2026
Industries · 6 min read
Custom vision models for industrial robots, trained on your own castings
The plant's own castings labeled from the cell's camera, defect, orientation and grasp point as the three questions, and a new variant handled by labeling it.
Andreas Ohrvall · Sep 30, 2026
Industries · 7 min read
Diabetes device inspection with a VLM and one detector across four device shapes
Four device shapes on one line: a question judges the faults nobody wrote a class for, a detector owns the count, and a new device generation is labeled first.
Rob Hickey · Sep 30, 2026
Edge · 6 min read
Edge AI fleet management for fifty runners across five plants
Fifty small boxes beside fifty recorders, one model version, and the IT patch that changes what every camera sends. What a fleet needs before the second plant.
Andreas Ohrvall · Sep 30, 2026
Operations · 6 min read
Face blurring for privacy in computer vision, done before the frame is stored
A blur applied after the request arrives is a blur applied to a frame that has already been copied. The step belongs at ingestion, beside the recorder.
Esdras Ntuyenabo · Sep 30, 2026
Industries · 7 min read
Fish size measurement on a grading line with a fixed camera
Fish boxed and sized against a calibrated channel, keypoints for true length when the tolerance demands it, and a bumped mount as the day every fish reads big.
Esdras Ntuyenabo · Sep 30, 2026
Operations · 6 min read
How to test computer vision model robustness before the weather does it for you
A vest model trained in June daylight meets October rain. Blur, darken, shift and compress the held-out frames first, and read recall per perturbation.
Rob Hickey · Sep 30, 2026
Operations · 6 min read
Machine vision lens selection worked back from the smallest defect in pixels
The lens is chosen from the defect, not from the drawer. Start with how many pixels a torn seal needs, and a later lens swap is a new camera to the model.
Finn Ellingwood · Sep 30, 2026
Industries · 6 min read
Medical device kit verification when the kit list changes every month
Each configuration is a checklist of parts, the camera boxes what is present and names what is absent, and a revision is a labeled part, not a rebuilt cell.
Ayman Quadir · Sep 30, 2026
Industries · 6 min read
Medical device label inspection reads the lot number only after finding the label
A detector finds the label, a question reads the crop, an implausible value goes to a person, and the two stages are scored apart so you know which to fix.
Rajiya Sultana · Sep 30, 2026
Edge · 6 min read
What computer vision model deployment means after the first week
Deploying a vision model is a decision about where it runs, how a new version replaces the old, and what tells you it has started to be wrong.
Andreas Ohrvall · Sep 30, 2026
Industries · 6 min read
An assembly line capacity monitor from a camera and a count, with no code written
Bottles counted in each sampled frame, a count above the line's normal as the rule, and the line lead told before the backup reaches the filler.
Ayman Quadir · Sep 30, 2026
Industries · 7 min read
Object dimension measurement from a fixed camera, using the mask rather than the bounding box
A rotated carton needs a polygon, scale comes from a printed reference or depth, the carrier's bands decide pass or fail, and a nudged mount breaks it.
Stephen Biswas · Sep 30, 2026
Industries · 6 min read
Object orientation from two keypoints, on a bottle line and a parcel sorter
Cap and base as keypoints, the angle between them as the answer, and consistent placement as the labeling rule that makes the whole thing hold.
Esdras Ntuyenabo · Sep 30, 2026
Industries · 7 min read
Parking lot occupancy detection from the camera already over the car park
A region per bay, occupied or empty as the call, a free-bay count for the sign and the manager, and snow, glare and a nudged lamp post as what goes wrong.
Finn Ellingwood · Sep 30, 2026
Industries · 6 min read
People counting and occupancy from the entrance camera, without a turnstile
A camera over the door counts who is inside right now from sampled frames. The Saturday crowd is where the count goes wrong, and the corrections show it.
Ayman Quadir · Sep 30, 2026
Industries · 7 min read
Plant height measurement with a camera and a stake of known height
Plant and stake as two classes, pixels to centimetres through the stake, a growth curve per plot from a question, and wind and a nudged mount absorbed by it.
Finn Ellingwood · Sep 30, 2026
Industries · 6 min read
Why the PlantDoc dataset is not your field, and what it is good for
A public leaf disease set is a photo of somebody else's field, lab backgrounds and box errors included. Use it as a baseline, then label the grower's frames.
Sheikh Srijon · Sep 30, 2026
Industries · 7 min read
Product colour inspection with a camera, measured inside the mask
A mask keeps the belt out, the colour is quantised to swatches and compared with a reference measured on the same camera, and a swapped camera moves everything.
Stephen Biswas · Sep 30, 2026
Industries · 6 min read
Product recognition AI at the till, checking the counter against the receipt
Items on the counter boxed and named by SKU, the set compared with the receipt, a mismatch sent to a person with the frame, and a new pack as what breaks it.
Ayman Quadir · Sep 30, 2026
Edge · 6 min read
Reducing computer vision inference costs when a store watches every aisle all day
The bill for watching a camera is set by how often you look, what you decode, and where the model runs. Most of a store's frames need no model at all.
Rajiya Sultana · Sep 30, 2026
Industries · 6 min read
Self-driving car perception and the white trailer against a bright sky
A model cannot find what its labels never showed it, the miss looks like nothing on the frame, and doubted fleet frames are how the training set catches up.
Rob Hickey · Sep 30, 2026
Industries · 7 min read
Tomato leaf disease detection from the greenhouse camera and the scout's phone
Blight, mould, spot and mite damage beside a healthy baseline, look-alike diseases merged rather than guessed, and a new season's light as what gets relabeled.
Sheikh Srijon · Sep 30, 2026
Operations · 6 min read
Underspecification in machine learning, and why two models with the same score do not both survive the second plant
Fifty training runs, one recipe, one validation score. The test set could not tell them apart, and the second plant's cameras can.
Rob Hickey · Sep 30, 2026
Industries · 7 min read
Wildfire smoke detection from a fixed ridge camera at dawn
Smoke is a class with no edges, cloud and valley mist are the false alarms to label, and a camera that never moves is what makes the first thin column visible.
Rob Hickey · Sep 30, 2026
Industries · 7 min read
Zero-shot pose estimation for a collaborative robot, and what makes it hold
A generalist pose model reads a worker's reach on day one, keypoints are smoothed across sampled frames, and the site's own reviewed frames make it hold.
Rob Hickey · Sep 30, 2026
Industries · 7 min read
Aerial fire detection from a drone patrol, smoke before the flame reaches the line
On a right-of-way patrol, smoke is a few dozen grey pixels that look like haze. Two boxed classes, an alert with a cooldown, and the bad-weather days kept.
Rob Hickey · Sep 29, 2026
Industries · 6 min read
AI in robotics after the robot ships, what the warehouse cameras keep learning
The forward camera boxed pallets and people well at the pilot site. Then the racking moved, and the edge cases the planner never saw came back for review.
Andreas Ohrvall · Sep 29, 2026
Industries · 6 min read
Automated sorting with computer vision, from the camera over the conveyor to the diverter
A box on every apple, a grade from the box, and an air jet that acts on it before the belt moves on. The new cultivar is when the model needs the graders again.
Rob Hickey · Sep 29, 2026
Labeling · 6 min read
COCO as a format you will use and a benchmark you should not trust
COCO JSON is the file a bottling line's labels travel in. The COCO benchmark is a score on somebody else's eighty classes, and none of them is a missing cap.
Sheikh Srijon · Sep 29, 2026
Computer vision · 6 min read
What the big vision models are for, and what runs on the camera
The big models find the frames, draw the first boxes and tighten them, and a person checks. The small model trained on those labels runs beside the recorder.
Andreas Ohrvall · Sep 29, 2026
Industries · 6 min read
Infrastructure asset management with computer vision from a truck at road speed
Signs, guardrails and poles pass the truck camera in a fraction of a second. The model finds them, the register says what should be there, a crew gets the gap.
Rob Hickey · Sep 29, 2026
Industries · 6 min read
Computer vision for office supply retail, empty pegboard hooks and the camera already watching them
The security camera at the end of the pen aisle sees every hook. A box per hook with an empty class, a rule with a cooldown, and a second store as a new site.
Ayman Quadir · Sep 29, 2026
Industries · 6 min read
CPG inventory cataloging with computer vision, every package named without a model per SKU
One detector boxes every package in the shelf photo, a question per crop reads brand, variety and size, and a new pack design is a spec change the queue shows.
Sheikh Srijon · Sep 29, 2026
Industries · 7 min read
Discrete vs process manufacturing, and where the camera earns its keep in each
Upstream the plant cooks a batch nobody can count. Downstream it fills jars anyone can. The camera does a different job on each side of the changeover.
Ayman Quadir · Sep 29, 2026
Labeling · 6 min read
EXIF orientation, the photo that is sideways only to the model
A phone photo looks upright on every screen and arrives rotated in training, because the pixels never turned. The check to run at import, before the first box.
Stephen Biswas · Sep 29, 2026
Labeling · 6 min read
What a model trained on ten frames is good for
Ten labeled frames from a new site camera give a first pass by the end of the afternoon. Trust it for the obvious, and let its doubts grow the set.
Sheikh Srijon · Sep 29, 2026
Labeling · 7 min read
Choosing an annotation tool for a month of inspection footage
The demo set labels itself in an afternoon. A month of drone footage is where the tool has to propose boxes, take corrections and review every label.
Sheikh Srijon · Sep 29, 2026
Industries · 7 min read
The smart factory on the floor, where the camera closes the loop
PLCs, OPC UA and the MES know what the machines did. The camera is the sensor that says what the part looks like, and it needs recalibrating like the rest.
Ayman Quadir · Sep 29, 2026
Industries · 6 min read
Two stages to read a plate, and why each needs its own score
A gate camera boxes the plate, then reads the characters. Score the two stages apart, or a rainy night's wrong read cannot tell you which one failed.
Rajiya Sultana · Sep 29, 2026
Industries · 6 min read
Manufacturing ERP systems and the gap between the ledger and the floor
The WIP module knows what somebody typed at shift end. A camera counting units off the fixture as they happen posts the record by webhook, minutes behind.
Andreas Ohrvall · Sep 29, 2026
Computer vision · 6 min read
Why a good detector can still be a bad counter
The queue camera finds every shopper on every frame and still gets the count wrong. Swaps, lost tracks and jitter are tracker mistakes, and review fixes one.
Stephen Biswas · Sep 29, 2026
Industries · 6 min read
Occupancy analytics with computer vision for the lot, the dock and the floor
Each bay is a polygon, each sampled frame says occupied or free, and the whiteboard becomes a chart by hour. Snow is what the correction rate shows first.
Esdras Ntuyenabo · Sep 29, 2026
Industries · 6 min read
OEE, OOE and TEEP, and the quality factor a camera can supply
The three metrics divide by different clocks and share one quality factor, estimated from a tray pulled at 2 pm. A camera counts every reject as it happens.
Ayman Quadir · Sep 29, 2026
Labeling · 6 min read
What an open-source annotation tool gives you and where it stops
Two years of fillet-line labels in a self-hosted tool, exported as CVAT XML, and no model watching the line. The tool was never the bottleneck. The loop was.
Ayman Quadir · Sep 29, 2026
Industries · 6 min read
Process compliance tracking with a camera, each step in order and for long enough
A model classes each step of a procedure per frame, a timer turns frames into a duration, and a rule fires when a step is skipped or cut short, frames attached.
Rajiya Sultana · Sep 29, 2026
Industries · 6 min read
Retail shelf item detection, a hundred products in one frame
A wide aisle camera sees a hundred boxes per frame, most of them identical neighbours. The rule is a tight box per facing, and the rare SKU stays in on purpose.
Sheikh Srijon · Sep 29, 2026
Operations · 6 min read
New footage lands in the bucket and shows up in the labeling queue
Packing-station cameras write to an S3 prefix, an event fires on each new object, and the frame is in the labeling queue minutes later, with a record of it.
Rajiya Sultana · Sep 29, 2026
Operations · 6 min read
Slicing the frame so the small things get found
A cracked insulator is a few dozen pixels in a frame thousands wide, and the detector shrinks the frame before it looks. Tiles keep the pixels, at a price.
Finn Ellingwood · Sep 29, 2026
Computer vision · 6 min read
What semantic segmentation labels and what it costs
Drivable path, corrosion area and crop rows are questions about pixels, and the answer is a class for every one. Painted labels cost more than boxes.
Esdras Ntuyenabo · Sep 29, 2026
Industries · 7 min read
Sensor fusion with computer vision, the tag says which pallet and the camera says what happened to it
The RFID portal reads a tag at the dock door. The camera above it sees the pallet lifted, filled or left behind, and catches the tag that fell off on the way.
Andreas Ohrvall · Sep 29, 2026
Industries · 6 min read
Solar panel detection in aerial imagery without counting the skylights
Over a suburb a panel is a few dozen pixels in a dense grid and a skylight is the same dark rectangle. Box per array, tile the frame, and keep the snow days.
Finn Ellingwood · Sep 29, 2026
Industries · 6 min read
Takt time vs cycle time, measured with the camera already over the station
Takt comes from demand and cycle time comes from the floor. A camera timing each unit through station 5 shows the gap per shift instead of once a quarter.
Ayman Quadir · Sep 29, 2026
Industries · 6 min read
Tire sidewall OCR, reading the DOT code off a black tire in a dark bay
Box the sidewall, light it from the side, read the DOT date code and size from the crop, and log it against the vehicle. A swapped bay camera is a new sensor.
Stephen Biswas · Sep 29, 2026
Industries · 6 min read
Training computer vision models on aerial imagery where a person is fifteen pixels tall
Over a corridor a crew member is a few dozen pixels. Tile instead of resizing, check orientation at import, label at full resolution, QA pass on every box.
Rajiya Sultana · Sep 29, 2026
Computer vision · 6 min read
What YOLO does in one pass and what that costs
One look at the frame is why a single-pass detector fits the box beside the recorder, and why it misses the small cluttered things a two-stage model finds.
Andreas Ohrvall · Sep 29, 2026
Labeling · 6 min read
AI labeled vs human labeled data, how much a model can do before a person has to look
On a shelf dataset the model's proposals are mostly right. On safety cones in a yard they are mostly wrong. Measure the gap, then keep a person on every label.
Finn Ellingwood · Sep 28, 2026
Labeling · 6 min read
Bounding boxes in computer vision, what a box teaches and what it reports
The same rectangle on a forklift is the lesson at training time and the answer at run time. Corner and centre formats, the shifted-box bug, floor in the box.
Esdras Ntuyenabo · Sep 28, 2026
Operations · 7 min read
Build or buy the layer that keeps a vision model accurate
Two engineers and a pilot can build a detector in a month. The review queue, the versioning and the rollout are what they are still building a year on.
Ayman Quadir · Sep 28, 2026
Operations · 7 min read
Benchmark the model on your own cameras before you believe a score
Two candidate models, two published scores, and a packaging line that only cares which one finds the torn label. The held-out week from your cameras decides.
Rob Hickey · Sep 28, 2026
Operations · 6 min read
Computer vision on the industrial HMI, the camera's verdict on the operator screen without the noise
The verdict as a state, the frame one click away, an alarm the operator acknowledges like any other, and everything else kept off the screen.
Andreas Ohrvall · Sep 28, 2026
Computer vision · 6 min read
When the outline is the answer and a box will not do
A weld defect whose area decides the rework and a part a robot has to grasp both need a mask. Here is what the mask costs to label and when a box is enough.
Esdras Ntuyenabo · Sep 28, 2026
Operations · 7 min read
A model trained on your warehouse beats one trained on the internet
A general detector calls the forklift a truck and the pallet nothing at all. The held-out week from your own cameras is the only score that counts.
Ayman Quadir · Sep 28, 2026
Computer vision · 6 min read
DETR explained, what the detection transformer changed about finding objects
Detection as set prediction, a fixed set of queries instead of anchors, no suppression step, and the small-object and slow-training problems later fixed.
Finn Ellingwood · Sep 28, 2026
Computer vision · 6 min read
Reading a training run before trusting the model
The loss curve says the run finished. The gap to validation, the recall on the crack class and the frames behind the score say whether the line can trust it.
Rob Hickey · Sep 28, 2026
Operations · 6 min read
The class that fails behind a good average, read off the confusion matrix
A shelf model with a healthy mAP and a dairy case full of unreported gaps. The matrix names the cell the failing class fell into, and the cell names the fix.
Rob Hickey · Sep 28, 2026
Labeling · 6 min read
Which words find the forklift, measured instead of guessed
Five ways to say forklift, five sets of boxes on aisle 6. Score each phrase on a small labeled set before Lexi labels the whole archive with the winner.
Sheikh Srijon · Sep 28, 2026
Operations · 6 min read
F1 score in computer vision, the failure that accuracy hides on the inspection line
Fifty porous welds in ten thousand parts make accuracy meaningless. F1 balances what was found against what was right, and the threshold is what moves it.
Rajiya Sultana · Sep 28, 2026
Labeling · 6 min read
How much training data a computer vision model needs from your own cameras
A weld defect detector got useful on a few hundred chosen frames. A sixty-class shelf model needed far more. The difference is the task and the variation.
Sheikh Srijon · Sep 28, 2026
Operations · 7 min read
Precision, recall and the mistakes each one lets through
A shelf model that cries wolf and an inspection model that waves the crack past are wrong in opposite ways. The threshold is the dial between them.
Rob Hickey · Sep 28, 2026
Labeling · 7 min read
How to label image data for computer vision so the model learns what you meant
Tight boxes on the forklift, the visible part of a hidden pallet, a written rule for the frame edge, and why a forgotten box teaches the model it is background.
Stephen Biswas · Sep 28, 2026
Labeling · 6 min read
Image annotation for robotics, when the robot will act on the label
In a bin-picking cell a box ten pixels loose is a missed grasp. Occlusion, blur, pose and the QA pass, written for the labels a gripper depends on.
Sheikh Srijon · Sep 28, 2026
Labeling · 6 min read
Labeling outdoor surveillance footage, the person forty pixels tall on the perimeter camera
Tight boxes on distant people and vehicles, occlusion labeled instead of skipped, classes the intrusion rule can use, and the rain and night frames kept in.
Sheikh Srijon · Sep 28, 2026
Industries · 7 min read
What a manufacturing execution system records and what only a camera can confirm
The MES logs a pass at station 12 because the operator pressed the button. The camera over the fixture is what says the cable was connected.
Ayman Quadir · Sep 28, 2026
Computer vision · 6 min read
What mAP measures and what it cannot tell you about your cameras
Mean average precision is built from per-class curves at an overlap line. A high score on the held-out set says nothing about the fence camera at 2 am.
Stephen Biswas · Sep 28, 2026
Labeling · 6 min read
Missing vs null annotations, an empty frame is a label and a forgotten box is a lie
An empty corridor, a defect-free shift and a stocked shelf are labels on purpose. A frame where someone forgot the box looks identical and teaches the opposite.
Rajiya Sultana · Sep 28, 2026
Labeling · 6 min read
Natural language image annotation, type what to look for and check what comes back
Hard hat and reflective vest, typed once and run across a site camera's frames. The boxes get checked one by one, and the lanyard clip gets drawn by hand.
Esdras Ntuyenabo · Sep 28, 2026
Labeling · 6 min read
Out-of-scope detection, what the model does when it sees something it was never taught
A screw detector on the phone line meets an operator's hand and calls it a screw. Null frames, a class for confusing things, a test set from the real camera.
Rob Hickey · Sep 28, 2026
Labeling · 6 min read
Promptable object detection, type the thing you want and get boxes back
Type forklift and the yard camera's frames come back boxed in seconds. The pallet jack and the distant hard hat are where the prompt stops and labeling starts.
Sheikh Srijon · Sep 28, 2026
Computer vision · 6 min read
Real-time object detection models, transformer or convolutional, and which size for the runner
A DETR-style detector drops the anchors and the suppression step. The size to run is decided by the box beside the camera, which the leaderboard cannot see.
Rob Hickey · Sep 28, 2026
Computer vision · 6 min read
ResNet-18 explained, what a residual backbone does underneath your detector
Why the shortcut connection made deep networks trainable, where the eighteen layers sit inside a detector on a plant camera, and why the frames matter more.
Stephen Biswas · Sep 28, 2026
Industries · 6 min read
Measuring a roof for solar from an aerial frame
Roof planes as polygons, a written rule for dormers, pixels to square metres through the ground sample distance, and the shadowed eaves sent back for review.
Ayman Quadir · Sep 28, 2026
Labeling · 6 min read
Video annotation, turning an hour of line footage into a training set
An hour from the bottling line is a hundred thousand near-identical frames. Sample every couple of seconds, label the ones that differ, track the rest.
Rajiya Sultana · Sep 28, 2026
Labeling · 7 min read
What active learning is, and why it labels the frames that change the model
Fifty thousand frames of good panels and nine splits. Active learning sends the frames the model doubts to a person first, and the cycle runs on the live line.
Rob Hickey · Sep 28, 2026
Computer vision · 7 min read
What an image embedding is and what it lets you find
A frame becomes a list of numbers where similar scenes sit close together. That finds the duplicates, the rare scene, and the night shift missing from your set.
Sheikh Srijon · Sep 28, 2026
Labeling · 6 min read
YOLO knowledge distillation, let the big model label and the small model run
A large model boxes the warehouse frames once, a floor lead corrects them, and a compact detector trained on that set runs beside the recorder.
Andreas Ohrvall · Sep 28, 2026
Operations · 7 min read
Active learning for computer vision on a line camera that never stops
The weld camera runs three shifts. The model returns the frames it doubts, the inspector corrects them, and past the threshold a new version trains and ships.
Rob Hickey · Sep 27, 2026
Edge · 8 min read
AI cameras vs IP cameras, and what changes when the model moves to the edge
The dock camera has streamed to a recorder for six years. A model can watch that stream in the cloud, on your servers or beside the recorder. No new camera.
Andreas Ohrvall · Sep 27, 2026
Operations · 7 min read
Camera focus measurement for a fixed camera that slowly goes soft
A lens loosened by vibration fails over weeks, and the model suffers before anyone sees blur. A sharpness score against the camera's own history catches it.
Rajiya Sultana · Sep 27, 2026
Operations · 6 min read
Danger zone monitoring with object detection on a site camera
A polygon over the crane swing radius on a site camera. People and vehicles as classes, the bottom of the box as the test, the alert with the frame attached.
Esdras Ntuyenabo · Sep 27, 2026
Edge · 7 min read
Cloud vs on-device inference for computer vision, and why the answer is usually both
A remote substation on a thin link and a plant with a footage policy. The runner decides on site, only doubted frames leave, the cloud trains the next version.
Andreas Ohrvall · Sep 27, 2026
Operations · 8 min read
Collecting image training data from a production line without keeping every frame
A carton line's model starts missing a new print. Sample the frames it doubts, drop near-duplicates, review them, and let corrections train the next version.
Sheikh Srijon · Sep 27, 2026
Edge · 8 min read
Running computer vision on the IP cameras a site already has
A store, a plant and a yard already stream RTSP to a recorder. That stream is the only requirement; the camera quote most projects begin with was never needed.
Andreas Ohrvall · Sep 27, 2026
Operations · 7 min read
Closing the computer vision feedback loop from an operator's flag to next month's model
The quality engineer on the cell line saw the false pass and had nowhere to put it. A flag on the frame, an event with the lot on it, and a queue that counts.
Rajiya Sultana · Sep 27, 2026
Operations · 6 min read
Three reasons a computer vision pilot never makes it from the lab to the floor
The cameras were fitted for security, the model was trained on someone else's parts, and drift arrived a month after launch unwatched. Each has a fix.
Ayman Quadir · Sep 27, 2026
Operations · 7 min read
Writing a computer vision problem statement before anyone labels a frame
The request said count the items on the shelf. The classes, the geometry, the action and the target were still missing, and each one changes what gets labeled.
Sheikh Srijon · Sep 27, 2026
Operations · 7 min read
A computer vision maturity framework for moving past the pilot, in five stages
Five plants, five pilots, no way to say which is further along. Five stages from a model on a bench to an operation nobody restarts, and what moves one up.
Ayman Quadir · Sep 27, 2026
Operations · 6 min read
A computer vision proof of concept that answers one question and then stops
A defect POC on stamped parts with one objective in numbers, frames from the real station, a defined endpoint, and the move to a pilot on the live line.
Ayman Quadir · Sep 27, 2026
Operations · 7 min read
Twenty questions to put in a computer vision RFP for a plant or a utility
Ask what happens to the frames the model doubts, how a new version reaches the cameras, whether footage can stay on site, and what the second substation costs.
Ayman Quadir · Sep 27, 2026
Edge · 7 min read
Running computer vision on RTSP camera streams from the recorder you already have
A dozen IP cameras and one NVR. One consumer per stream, a frame about every two seconds, TCP when the switch is busy, and a bitrate change ruled out first.
Finn Ellingwood · Sep 27, 2026
Edge · 6 min read
Deploying computer vision models offline, where the network does not reach
A vessel at sea for three weeks and a plant with no outbound video: the runner beside the recorder, alerts on site, doubted frames leaving only on a link.
Andreas Ohrvall · Sep 27, 2026
Operations · 8 min read
Dwell time detection in video, measuring how long rather than how many
A shopper at the dairy case, a worker inside a guard, a truck at a dock. Dwell needs identity across frames, a zone with a soft edge, and stream timestamps.
Stephen Biswas · Sep 27, 2026
Operations · 7 min read
Event-driven object detection, turning a bounding box into an alert someone acts on
The packing area camera sees cartons all day. Three stacked is the rule, the frame goes to the supervisor, and a cooldown makes one stack one alert.
Esdras Ntuyenabo · Sep 27, 2026
Operations · 7 min read
How to improve computer vision model accuracy by starting with one defect class
A coffee bag line wanted eight defect classes and got none. Torn bag first, frames from the station camera, a written guideline, then the loop adds the rest.
Rob Hickey · Sep 27, 2026
Operations · 6 min read
Human-in-the-loop computer vision, the mistakes the model is sure about and the inspector who catches them
A quality inspector flags a wrong detection in seconds, the confidently wrong frames no threshold would catch, and the correction rate as the signal.
Rob Hickey · Sep 27, 2026
Operations · 6 min read
Monitoring a computer vision model in production, and what to do when the corrections start climbing
A shelf model meets the Christmas displays: the correction rate as the signal, doubted frames first in the queue, a new version compared on the same frames.
Rob Hickey · Sep 27, 2026
Operations · 8 min read
Monitoring data drift in computer vision, and the one signal that catches both kinds
A container yard camera at night is data drift. A damage policy that now counts dents is concept drift. Only the correction rate catches both.
Rob Hickey · Sep 27, 2026
Edge · 7 min read
On-premise computer vision inference when the footage cannot leave the fence
A utility, a plant and a vessel whose security teams refuse outbound video. The model runs on your servers or beside the recorder; only doubted frames leave.
Andreas Ohrvall · Sep 27, 2026
Operations · 6 min read
PLC and computer vision integration, where the model fits next to the controller
How a scan cycle works, why the PLC keeps the reject arm, the detection arriving as a webhook into the broker the PLC reads, and the latency a line can absorb.
Andreas Ohrvall · Sep 27, 2026
Operations · 6 min read
Presence and absence detection, is the barrier in place and how long has it been gone
A forklift bay, a shadow board and a safety gate as present or absent per frame, a rule that fires after four minutes gone, and the pallet in the way.
Stephen Biswas · Sep 27, 2026
Operations · 7 min read
Production deployment is the halfway point of a computer vision model's life
A packaging line adds a flavour and a store resets for autumn. The model that scored well in evaluation meets both on a Monday, and the second half begins.
Rob Hickey · Sep 27, 2026
Operations · 6 min read
Scaling a computer vision operating model from one camera on one line to twenty across five plants
One metric, time from request to live inference, the pilot's threshold written down, and every new plant treated as a new site with its own labeled window.
Ayman Quadir · Sep 27, 2026
Operations · 6 min read
Video analytics with vision AI, the three stages every camera system shares
From a motion rule that fires at dusk to a model that answers whether the package is damaged, and where the runner and the review queue sit in it.
Andreas Ohrvall · Sep 27, 2026
Operations · 6 min read
Video process monitoring, turning a conveyor camera into a count you can trust
A conveyor camera counting parts and a dock camera counting pallets: track so each is counted once, write the crossing rule as a sentence, watch the volume.
Rajiya Sultana · Sep 27, 2026
Operations · 7 min read
Ten questions that tell you whether a vision AI pilot will ever leave the pilot line
A pilot with no owner, no definition of done and no path for the frames it gets wrong stays a pilot. Ten questions a plant committee can score in one sitting.
Ayman Quadir · Sep 27, 2026
Operations · 6 min read
Choosing a vision AI vendor, buying a capability or renting one forever
The first site is a project either way. The tell is whether the second is a statement of work or a window of frames, and whose people correct labels later.
Ayman Quadir · Sep 27, 2026
Industries · 6 min read
AI crop analysis in the greenhouse, catching tomato disease before it spreads down the row
Lesions boxed by disease with a healthy class, a question of the footage about how far a patch spread since last week, and a model that turns with the season.
Rob Hickey · Sep 26, 2026
Industries · 6 min read
Ceramic defect detection for hairline cracks a fixed-rule camera cannot learn
Edge chips, hairline cracks and pinholes on the tile line after the kiln, masks where the area sets the grade, and the new glaze as the day to relabel.
Esdras Ntuyenabo · Sep 26, 2026
Industries · 6 min read
Chocolate box inspection with a camera over the tray line
Each piece boxed by type, a plain check against the box template, damage as its own class, and the new spring assortment as the day the labels go stale.
Stephen Biswas · Sep 26, 2026
Industries · 7 min read
Computer vision inventory monitoring from a frame every few minutes
A shelf camera sampled every few minutes, each facing boxed and counted, a record per detection, and an alert when a facing stays empty past the hour.
Sheikh Srijon · Sep 26, 2026
Industries · 6 min read
Error proofing in manufacturing with object detection, at the station where a limit switch cannot judge
A sensor confirms the clip is present. The camera says it is backwards. Every verdict kept with its frame, the threshold retuned when the mistake rate moves.
Ayman Quadir · Sep 26, 2026
Industries · 7 min read
What a station camera takes off the manufacturing operator's plate
Presence, orientation, fasteners and the label, checked by the camera on every part, so the operator keeps the judgment calls and the exceptions.
Ayman Quadir · Sep 26, 2026
Industries · 6 min read
Pipeline inspection with camera nodes that watch for the crack before it is a leak
Cameras on the exposed spans, classes written with the integrity team, the alert raised at the node with the frame, and rain showing in the corrections first.
Andreas Ohrvall · Sep 26, 2026
Operations · 6 min read
Counting objects on a conveyor belt without counting any of them twice
Bolts tracked across frames so each is counted once, touching parts as the label to settle, and a compression change as the step that breaks the count.
Finn Ellingwood · Sep 26, 2026
Industries · 7 min read
Dimensional defect inspection at the press outfeed, for the bracket that looks perfect and will not fit
Hole spacing, gap width and alignment measured on every part against tolerance, a drifting tool offset caught as a slope, and a new die labeled before the run.
Finn Ellingwood · Sep 26, 2026
Industries · 6 min read
Dock door utilization tracking from the yard camera
Trailers boxed, each tied to a door, occupied or empty with a dwell time per door, and the rainy mornings when the door numbers go unread.
Rajiya Sultana · Sep 26, 2026
Industries · 6 min read
Expiration date detection on a curved bottle neck at the packing station
The code found as a box on the neck, read as a string, checked against the plant's own format, and the case held on a mismatch. A new coder is a spec change.
Esdras Ntuyenabo · Sep 26, 2026
Industries · 6 min read
FDA food label compliance with a camera over the labeling station
Box the nutrition panel and the allergen statement, read them, compare them with the spec, and hold the case on a mismatch before the sealer.
Ayman Quadir · Sep 26, 2026
Industries · 6 min read
First pass yield from a station camera, and the hidden factory it exposes
Final yield says nearly everything shipped. The rework bench says otherwise. A pass, review or fail on every part at every station makes the gap visible.
Rajiya Sultana · Sep 26, 2026
Industries · 6 min read
Instance segmentation for robotic manipulation from twenty-five labeled frames
A first small batch, the camera angle and the label scheme checked before anything scales, and masks the robot can close a gripper on.
Sheikh Srijon · Sep 26, 2026
Industries · 6 min read
Juice box quality inspection at the straw applicator, with a tag on every carton
Acceptable, missing, bent or pierced on every carton from the line's own frames, and why a nudged camera mount is the failure a classifier shows first.
Rob Hickey · Sep 26, 2026
Industries · 7 min read
Lot code and expiry date verification with vision AI before the pouch moves on
The strip found on every pouch wherever the film pulled it, the crop read, format and date logic checked, and a failed read stopping the pouch, not guessing.
Stephen Biswas · Sep 26, 2026
Industries · 7 min read
OCR in manufacturing for the date codes and serials the printer did not make easy
Dot-matrix dates on cardboard, label text and etched serials on steel, each found by a detector, checked against the schedule, sent to a person when smudged.
Rajiya Sultana · Sep 26, 2026
Industries · 6 min read
Optical character verification checks the print against the work order instead of reading it from scratch
The expected lot and expiry come from the order, a detector finds the print on every pack, and the pack is held when they disagree. A new layout is a relabel.
Stephen Biswas · Sep 26, 2026
Industries · 6 min read
Paint defect detection after the booth, with object detection tied to the process cause
Orange peel, runs, dirt and solvent pop as classes, each tied to what the booth would change, so the alert says what to fix. A tighter standard is a relabel.
Sheikh Srijon · Sep 26, 2026
Industries · 7 min read
PCB defect detection on bare boards, with a severity on every box
Opens, shorts and mouse bites boxed on the bare panel, critical ones rejected, repairable ones sent to rework with the frame, a new revision labeled early.
Esdras Ntuyenabo · Sep 26, 2026
Industries · 6 min read
Pipe and tube defect detection with object detection, and a rule that gives every metre a verdict
Pinholes, cracks and pitting boxed on the mill's own tube, a threshold in the rule where the inspector can read it, and the borderline metre reviewed.
Rajiya Sultana · Sep 26, 2026
Industries · 7 min read
Predictive maintenance with computer vision on the bearing that is about to seize
A camera on the bearing housing, the belt and the pipe joint sees the weep, the fray and the loose bolt weeks before the stop, and sends the frame.
Rob Hickey · Sep 26, 2026
Industries · 7 min read
Rebar counting from one photo of the bundle end, with a bounding box on every bar and a record for the ticket
A camera over the saw end of the bundle, a box on every bar, the count checked against the delivery ticket, and a re-mounted camera as the drift to expect.
Finn Ellingwood · Sep 26, 2026
Industries · 6 min read
Robotic pick and pack with a bounding box and a mask on every item, and why the arm still misses
A box finds the item, a mask finds where the cup should land, foil packaging punches holes in the depth map, and a nudged camera mount shows up as failed picks.
Rob Hickey · Sep 26, 2026
Industries · 7 min read
Traditional sensors vs AI cameras on the line, and why the camera answers the questions the eye cannot
The photoelectric eye at the capper counts every bottle and cannot see a crooked cap. The camera can, and its verdict reaches the PLC without a wire.
Ayman Quadir · Sep 26, 2026
Industries · 6 min read
Vehicle damage localization with instance segmentation, so the estimate names the panel and the area on it
One mask for the dent and one for the panel it sits on, from the claimant's own phone photos, so the estimate says bumper and how much of it.
Sheikh Srijon · Sep 26, 2026
Industries · 6 min read
Visual assembly verification with a golden image per variant
One golden image per variant at a shared station. A missing fastener, a warped housing or a wrong cable route is flagged as a deviation, no defect list needed.
Rajiya Sultana · Sep 26, 2026
Industries · 6 min read
A visual quality management system that learns from every flagged part
A scratch flagged on line two at 3 am, the frame to the shift lead, the record tied to the batch, and the corrected label training the next version.
Rajiya Sultana · Sep 26, 2026
Industries · 6 min read
Warehouse rack occupancy detection from a camera on the forklift, so an empty bay is a finding and never a guess
Rack slots and pallets as separate classes, occupancy written back per position with the frame, and stock staged in the aisle as the first failure to rule out.
Andreas Ohrvall · Sep 26, 2026
Industries · 6 min read
Wood defect detection on the grading line, with a bounding box on every knot
Knots live or dead, checks kept apart from splits, resin pockets counted, and the board routed by a rule the mill can read. The grader keeps the borderline.
Ayman Quadir · Sep 26, 2026
Industries · 7 min read
Counting the screws in a kit with object detection over the bench
A camera over the kitting bench counts every screw and bottle one box at a time. Exact for kits, a band for cases, and a half-hidden item is a written rule.
Rajiya Sultana · Sep 25, 2026
Industries · 7 min read
Automated water meter reading with a camera in the vault
Ten digit classes in a fixed row turn a truck roll into a frame. A rolling digit and condensation on the glass are the frames that come back to a person.
Stephen Biswas · Sep 25, 2026
Industries · 7 min read
Body-in-white inspection with a station camera on unpainted steel
Dents on a bare body shell hide in the reflections. Masks give the extent, a second pass decides reportable, and a tightened tolerance is a spec change.
Finn Ellingwood · Sep 25, 2026
Industries · 7 min read
Computer vision in automotive manufacturing, from the body shop to final inspection
Welds in the body shop, runs in the paint shop, a missing clip at trim, on cycle times in seconds. A model-year change is the day the model needs new frames.
Ayman Quadir · Sep 25, 2026
Industries · 7 min read
Computer vision for autonomous mobile robots that pull pallets from dark trailers
A robot's forward cameras in a dark trailer find a worn pallet and its pose, keep people in view, and the second warehouse is a new site with its own frames.
Andreas Ohrvall · Sep 25, 2026
Industries · 7 min read
Computer vision in logistics, from the dock door to the sorter
A distribution centre's cameras already see the dock, the sorter and the aisles. Packaging refreshes, new SKUs and moved cameras are the part to plan for.
Ayman Quadir · Sep 25, 2026
Industries · 6 min read
Computer vision inventory management with a camera over the stockroom shelf
Stacks counted on sampled frames without a scan, discrepancies flagged against the ledger with the frame, and a packaging change as the relabel.
Sheikh Srijon · Sep 25, 2026
Industries · 6 min read
Object detection on satellite imagery finds every house inside the pipeline buffer
A proposed corridor over a basemap, houses and barns proposed as boxes, a GIS analyst approving each one, and summer leaf cover as the season that hides roofs.
Sheikh Srijon · Sep 25, 2026
Industries · 8 min read
Semiconductor wafer inspection with a bright-field camera at intake
A cracked wafer rejected at intake costs a wafer. Found after lithography it costs every step spent on it. Three classes, and masks where extent decides.
Finn Ellingwood · Sep 25, 2026
Industries · 8 min read
Computer vision in steel manufacturing, from the ladle shell to the coil
A thermal camera on the ladle stand sees the lining fail from the outside, and a replaced camera is the day the hotspot map has to be redrawn.
Rob Hickey · Sep 25, 2026
Industries · 7 min read
Contact lens defect inspection with pass, review and fail as three answers
A dark-field camera turns a crack in a clear lens into a bright line. The lens the model cannot decide goes to a person, and that middle bucket is the product.
Rob Hickey · Sep 25, 2026
Industries · 7 min read
Crack detection with computer vision as a tag, a box or a width in millimetres
A tag says whether a footing is cracked, a box says where, and only a mask gives a width the engineer can act on. The low winter sun turns texture into cracks.
Sheikh Srijon · Sep 25, 2026
Industries · 7 min read
How to detect small defects with computer vision when three resizes erase them
A hairline crack on a blade and a micro-scratch on a display vanish before the first layer sees them. Tiling keeps their pixels; the labeler sees them first.
Rob Hickey · Sep 25, 2026
Industries · 6 min read
Drone structural damage detection with cracks and spalling confirmed by the engineer
Masks for cracks and spalling from a close drone pass, so area compares pass to pass. The engineer confirms each flag, and a re-planned flight moves every mask.
Rob Hickey · Sep 25, 2026
Industries · 7 min read
Fall detection with pose estimation on the warehouse cameras a plant already has
A worker down behind the racking at 2 am, a torso line that drops between samples, and an alert with the frame attached that reaches someone who can walk there.
Rob Hickey · Sep 25, 2026
Industries · 7 min read
Flange defect inspection with a bounding box on every pinhole in the sealing face
A pinhole on a sealing face is a leak in service. The camera boxes scratch, crack, dent and pinhole, and the flange it cannot decide goes to a person.
Rajiya Sultana · Sep 25, 2026
Industries · 6 min read
Flaw detection with computer vision for the flaws an inspector stops seeing at hour eight
Surface, structural and material flaws on one welded bracket, why inspectors disagree by hour eight, and the review loop that holds the 7 am standard.
Rob Hickey · Sep 25, 2026
Industries · 8 min read
Georeferencing drone detections from a box in the frame to a GPS point the crew can drive to
Heading, altitude and field of view turn a boxed insulator into a coordinate. The log and the video disagree on time, and a re-planned flight moves every box.
Andreas Ohrvall · Sep 25, 2026
Industries · 7 min read
Hog ring inspection on the seat line with object detection under the fabric
Every hog ring boxed and its closure read among wires and clips, the count checked against the build order, and the ring under the fabric is a labeling rule.
Esdras Ntuyenabo · Sep 25, 2026
Industries · 7 min read
IV bag fill level and leak detection on a transparent bag with a line camera
Fluid and ports find the bag, fill level is a class, and whether a port is wet is a question put to the frame. A swapped camera moves every highlight.
Stephen Biswas · Sep 25, 2026
Industries · 6 min read
Six Sigma with a camera on the line for the measure and analyze phases
Defect counts and cycle observations from the station camera instead of a sample, and a Control phase that treats a moving base rate as a finding.
Ayman Quadir · Sep 25, 2026
Industries · 7 min read
Manual assembly QA with a camera that catches the wrong part in the assembler's hand
A camera over the bench boxes and names the part in hand, checks it against the next step, and lights a lamp before the next screw goes in.
Finn Ellingwood · Sep 25, 2026
Industries · 7 min read
Object detection on aerial imagery and why the ground detector fails at four hundred feet
Insulators a few dozen pixels wide, pipelines at every angle, solar tables packed edge to edge, and why a re-planned flight path is a camera that moved.
Stephen Biswas · Sep 25, 2026
Industries · 6 min read
Oil spill detection from a drone, mapped by thickness so the booms go to the right place
Sheen, rainbow and true colour as masks on the drone frames, in the appearance codes responders already use, so the map says where the booms go.
Sheikh Srijon · Sep 25, 2026
Industries · 6 min read
Pallet scanning with a camera portal that reads the label from whichever side it landed on
Cameras on four sides of the dock door find the label wherever the wrapper left it, decode the barcode, read the text, and log every scan with its frame.
Andreas Ohrvall · Sep 25, 2026
Industries · 7 min read
Retail planogram compliance checked from one shelf photo on the aisle camera
Right product, right facing, right price, from one frame of the cereal bay. Products and tags boxed, rows parsed, prices read, and a reset is a spec change.
Ayman Quadir · Sep 25, 2026
Industries · 7 min read
Robotic welding defect detection on every weld the robot lays down
A camera at the weld cell names porosity, undercut and missing fusion on every part, sends the borderline weld to a person, and a new torch needs new frames.
Esdras Ntuyenabo · Sep 25, 2026
Industries · 8 min read
Surface defect detection in manufacturing, from the camera to the grade
Scratches, pits and stains under a controlled light become boxes, the boxes become a grade by rules, and a shifted defect mix is a threshold to revisit.
Sheikh Srijon · Sep 25, 2026
Industries · 7 min read
Tablet defect inspection that says which defect, so the engineer knows which press setting to check
Every tablet boxed, each crop classed as capping, lamination, chip or crack, and the class says which press setting to check. A new defect is its first labels.
Sheikh Srijon · Sep 25, 2026
Industries · 7 min read
Verify label placement on packages before the sorter with a camera at induction
Package and label boxed, the label checked against the zone the scanner can read, skew and cover flagged with the frame. A new carton size is a spec change.
Rajiya Sultana · Sep 25, 2026
Industries · 6 min read
AI visual inspection as the nondestructive testing step a camera can take over
Visual testing is the first NDT gate, its acceptance criteria are already written, and a camera can apply them to every weld instead of one in twenty.
Rob Hickey · Sep 24, 2026
Industries · 6 min read
Appearance inspection systems that judge scratches, chips and burrs the same way on every shift
The station, the light and the written standard matter more than the model. The outlines carry the limit, and a tightened tolerance makes every label wrong.
Rajiya Sultana · Sep 24, 2026
Industries · 7 min read
Automated pallet accounting from the camera over the staging zone
A polygon on the frame, every pallet tracked so it is counted once, entries and exits as the ledger, and a wash-down that nudges the camera as the failure.
Andreas Ohrvall · Sep 24, 2026
Industries · 7 min read
Bottle cap inspection with a camera at the capper outfeed
Sealed, loose and missing as three classes, a reject on the second two before the case packer, and the loose rate per batch as the number the technician wants.
Esdras Ntuyenabo · Sep 24, 2026
Industries · 6 min read
Coffee bean inspection with computer vision on the roasted bean belt
A camera over the roasted bean conveyor finds stones, quakers and foreign material as masks. A new origin lot is the day the defect rate moves.
Stephen Biswas · Sep 24, 2026
Industries · 6 min read
Computer vision pill inspection catches one damaged tablet in a thousand
A camera over the tablet conveyor before the blister packer, recall set on the defect side, and thresholds re-tuned when a new batch moves the base rate.
Rob Hickey · Sep 24, 2026
Industries · 7 min read
Retail customer movement monitoring from the cameras already on the ceiling
People boxed and tracked, zones drawn as polygons, dwell and occupancy per zone, and the camera nudged during a Sunday reset as what moves every zone at once.
Ayman Quadir · Sep 24, 2026
Industries · 7 min read
Computer vision for robotics, what the cameras have to answer before the arm moves
Detection, segmentation, pose estimation and depth, in that order, with the label each one needs, and a firmware update as the first failure to check.
Andreas Ohrvall · Sep 24, 2026
Industries · 6 min read
Machine vision or computer vision on the line, and when each breaks
A threshold rule that died with the old fluorescent tubes, a template that broke on the new variant, and the model that is corrected rather than re-engineered.
Rob Hickey · Sep 24, 2026
Industries · 7 min read
Computer vision for workplace safety, six things one camera can watch for
Person, PPE, vehicle, zone, posture and spill from one yard camera, the person-then-PPE design, why a stock model calls a cap a hard hat, alerts to the lead.
Sheikh Srijon · Sep 24, 2026
Industries · 7 min read
Container yard management with computer vision, knowing where every box sits without a radio call
Row cameras and a hostler camera box each container and its ID panel, vote the number across a pass, and answer where the reefer went last night.
Stephen Biswas · Sep 24, 2026
Industries · 7 min read
Corrosion detection with computer vision that measures rust as an area
A tag to triage the survey, a box to send the crew, a mask to say whether the patch grew since last quarter. Wet steel is where the model struggles.
Sheikh Srijon · Sep 24, 2026
Industries · 7 min read
Cycle time measured per part by the camera over the cell
One camera over a machining cell gives every housing its own clock, shows where it waited, and a nudged bracket is what quietly moves every zone.
Ayman Quadir · Sep 24, 2026
Industries · 7 min read
Defect inspection in manufacturing, where on the line each kind of defect is caught
Surface, dimensional and assembly defects on one bottling line, the camera that catches each, and the review that holds the standard across shifts.
Ayman Quadir · Sep 24, 2026
Industries · 7 min read
How to detect metal defects with computer vision, one model across brushed and polished parts
Scratch, dent, crack and pit on a stamping line, the light that shows each, a decision layer that knows the cosmetic face, a new alloy labeled in advance.
Esdras Ntuyenabo · Sep 24, 2026
Industries · 6 min read
How to detect solar panel failure with computer vision, snow, soiling and hot spots panel by panel
Each module as its own outline from a drone pass, snow and soiling on the visible frames, hot spots on the thermal ones, an alert per string to the O&M tech.
Sheikh Srijon · Sep 24, 2026
Industries · 7 min read
Injection molding defect detection on syringe barrels at the mold outfeed
Cracks, short shots, flash and black specks boxed on every barrel as it leaves the press, logged to the lot, and a resin change is what dates the labels.
Ayman Quadir · Sep 24, 2026
Industries · 7 min read
Label inspection with computer vision, six checks a carton passes before the case is sealed
Presence, placement, orientation, print, content and barcode, each check cropping the frame for the next, from the camera on the labeler outfeed.
Esdras Ntuyenabo · Sep 24, 2026
Industries · 7 min read
Lights-out manufacturing with a camera on the chip pile and the part seat
An unattended machining cell from 10 pm to 6 am, three cameras where an operator's eyes used to be, and the frame that reaches the on-call phone in time.
Andreas Ohrvall · Sep 24, 2026
Industries · 6 min read
Machine guarding with computer vision, and the steel guard it does not replace
A polygon over the point of operation, a person inside it as the rule, a warning the guard cannot give. A wiped lens moves the zone without anyone noticing.
Ayman Quadir · Sep 24, 2026
Industries · 6 min read
Object alignment detection for the label, the connector and the gasket on one station
Is it where it should be, and turned the right way. A box answers most of it, pose answers the last two degrees, and the reference moves when the camera does.
Stephen Biswas · Sep 24, 2026
Industries · 6 min read
Object detection on a food line when no two items look alike
Seals, foreign material and portion counts from one camera over a fillet line, the skylight as the drift the corrections catch, and a new recipe labeled early.
Ayman Quadir · Sep 24, 2026
Industries · 6 min read
On-shelf availability monitoring with the aisle camera a store already has
Empty facings boxed on the bread bay, glare and black packaging as the false alarms, a cooldown rule to the restocker, alert volume watched against itself.
Rob Hickey · Sep 24, 2026
Industries · 7 min read
Training a model for a defect that appears three times a month
A hairline crack on a board keeps its pixels through tiling, every rare frame is kept and weighted, and the doubted frames are how the class grows.
Sheikh Srijon · Sep 24, 2026
Industries · 6 min read
Steel strip defect inspection with pass, review or fail for every metre
A camera over the finishing line boxes scratches, inclusions, crazing and scale, sends uncertain metres to the inspector, and reads the mix off the corrections.
Rajiya Sultana · Sep 24, 2026
Industries · 6 min read
Surface defect detection on machined medical parts, with the scratch boxed where it is
A camera at the deburring station puts a box on the burr, the crack and the pit so the inspector sees where, and a new finish spec is the change to label.
Rajiya Sultana · Sep 24, 2026
Industries · 6 min read
Telecom tower inspection for corrosion, measured as area from a drone pass up the legs
Rust on a lattice tower as a mask so two flights can be compared, the member it sits on as the priority, and salt haze as the frames the model has seen least.
Finn Ellingwood · Sep 24, 2026
Industries · 7 min read
Thermal drone solar panel hotspot detection, and the palette that fools the model
Hotspots as boxes against a uniform module, white-hot against ironbow as what changes the pixels, the thermographer confirming, a new camera as a new baseline.
Stephen Biswas · Sep 24, 2026
Industries · 6 min read
Vision AI and the eQMS, so the quality record starts with a frame instead of a typed note
The nonconformance module knows what an operator typed at the end of the shift. A line camera raises it with the frame, the lot and the time attached.
Rajiya Sultana · Sep 24, 2026
Industries · 6 min read
Vision guided robotics, giving the robot a reason to look before it moves
A casting a few millimetres off its nest stops a blind program. The camera-guided one finds the part, estimates its pose and adjusts, until the mount is bumped.
Finn Ellingwood · Sep 24, 2026
Labeling · 6 min read
Automated image labeling, where a model proposes and a person makes it true
Type what to look for, let the model box every frame of the right-of-way, then check each label before it trains. Proposals are drafts until a person agrees.
Sheikh Srijon · Sep 23, 2026
Labeling · 6 min read
Writing a class ontology for defect labeling before the first box
On a weld line, the boundary, occlusion and inclusion rules come before the first box. When the tolerance tightens, the old labels are wrong.
Rob Hickey · Sep 23, 2026
Industries · 8 min read
Computer vision for construction site safety, a warning before the worker and the excavator meet
A pole camera boxes people and machines, draws a danger zone that moves with the excavator, and sends the frame when someone walks into it.
Ayman Quadir · Sep 23, 2026
Industries · 7 min read
Computer vision in agriculture, from the sprayer boom to the packing line
Weeds against beet rows at dawn, lesions on a leaf, bruises on a packhouse belt, and labels that go stale as the season turns.
Ayman Quadir · Sep 23, 2026
Industries · 6 min read
Computer vision applications on a factory floor, four jobs for the cameras already there
Defect detection, assembly verification, safety and inventory on one plant's cameras, with cosmetic against functional written into the labeling schema.
Ayman Quadir · Sep 23, 2026
Operations · 6 min read
Turning a detection into an OPC UA tag the PLC can read
A count, a presence flag and a confidence value as OPC UA variables, written on change from the alert webhook, while the reject decision stays with the PLC.
Stephen Biswas · Sep 23, 2026
Industries · 6 min read
Defect detection on the line, why the tolerance decides whether you draw a box or a mask
Two scratches on a stamped panel, one under the tolerance and one over, look the same inside a box. The geometry follows the disposition rule.
Sheikh Srijon · Sep 23, 2026
Industries · 6 min read
Glass defect detection with computer vision, scratches where the background shows through
On a bottle line the conveyor shows through the glass and a scratch and a reflection are the same bright line. The labels have to follow the light.
Esdras Ntuyenabo · Sep 23, 2026
Operations · 7 min read
Integrating machine vision with PLC, SCADA, MES and ERP systems
A cracked weld is a bit for the PLC, a thumbnail for the HMI, a record for the MES and a yield figure for the ERP. Shape it once, beside the recorder.
Andreas Ohrvall · Sep 23, 2026
Industries · 6 min read
Choosing a machine vision system for visual inspection on the line
Backlight, ring, coaxial or dome for the defect you need to see, why a badly lit frame cannot be fixed later, and vendor questions about the second year.
Stephen Biswas · Sep 23, 2026
Operations · 7 min read
Getting a defect detection to the PLC over MQTT
One webhook from the alert into the plant's broker, a topic per line and camera, and the PLC and the SCADA screen subscribe like any other client.
Andreas Ohrvall · Sep 23, 2026
Industries · 6 min read
Package damage detection with a camera at the packing station
A camera at the packing station, five classes of damage, and an alert with the frame attached so the station lead pulls the box before it ships.
Sheikh Srijon · Sep 23, 2026
Industries · 6 min read
Part inspection with computer vision, the six questions a station camera answers
Present, how many, which way round, what colour, damaged and aligned each need a different label. A component revision means labeling the new part first.
Rajiya Sultana · Sep 23, 2026
Industries · 6 min read
PPE detection in production, deciding who is wearing the hard hat
A gate camera, three classes and a spatial rule decide compliance per person. The hard hat carried in a hand is the label that decides whether the model holds.
Finn Ellingwood · Sep 23, 2026
Computer vision · 5 min read
Pretrained vs fine-tuned detection models, when the stock detector cannot see the thing you need
A stock detector sees every worker on the scaffold and has no hard hat class, and no threshold will invent one. The site's own frames are what add it.
Esdras Ntuyenabo · Sep 23, 2026
Industries · 6 min read
Preventing food recalls with a camera at the labeller and one over the belt
The label matched to the run, the allergen statement checked for presence and legibility, foreign objects boxed in the tray, a recipe change as a spec change.
Rajiya Sultana · Sep 23, 2026
Industries · 6 min read
Production line monitoring with the cameras already over the line
The cameras were installed for disputes. A model turns a missing cap, a low fill and a count into events for the line lead. The mount decides whether it holds.
Andreas Ohrvall · Sep 23, 2026
Industries · 8 min read
Reading analog gauges with a camera, without replacing them
A camera on the dial turns a clipboard round into a stream of readings, and a nudged bracket is what quietly breaks it.
Rob Hickey · Sep 23, 2026
Industries · 6 min read
Reducing manufacturing scrap with a camera on the line, because scrap is a timing problem
A torch drifts at 9 am and the shift finds out at 2 pm. Scrap scales with that lag, and the camera flags the first bad bead rather than the hundredth.
Rob Hickey · Sep 23, 2026
Industries · 6 min read
Red zone monitoring with a camera over the forklift lane
Person and vehicle classes trained on the site's own frames, the zone as a polygon, dwell as the rule, and a nudged camera as what silently moves the zone.
Rob Hickey · Sep 23, 2026
Operations · 6 min read
Getting a defect record into the MES without touching the PLC
The alert webhook posts a quality record to the plant's REST endpoint, the work order is updated with the frame, and the reject arm keeps listening to the PLC.
Rajiya Sultana · Sep 23, 2026
Industries · 6 min read
Retail queue analytics with the camera over the tills, measuring the wait rather than the crowd
Each shopper is tracked from joining the queue to reaching a till, the alert is minutes above target, and sun through the front windows breaks it first.
Ayman Quadir · Sep 23, 2026
Industries · 6 min read
Security camera monitoring with computer vision, turning the camera wall from a recorder into a watcher
A model on the existing cameras turns entry, lingering, a bag left behind and a gate queue into alerts with the frame. Night and rain are what it learns second.
Ayman Quadir · Sep 23, 2026
Industries · 7 min read
Shelf price verification with the aisle cameras a store already has
Box every shelf label, read the price, compare it with the till, and alert with the frame attached. A promotion change is a spec change for the rule.
Ayman Quadir · Sep 23, 2026
Operations · 6 min read
Deploying a computer vision model to a second plant, beyond the weights file
What the second plant needs beyond a weights file: recorded thresholds and the test set behind them, the alert template, acceptance criteria, its own frames.
Andreas Ohrvall · Sep 23, 2026
Industries · 7 min read
Transmission line inspection with object detection from a survey flight
Damaged strands, broken insulators, foreign objects and vegetation from one flight, tiled so a strand keeps its pixels, with recall set per hazard.
Rob Hickey · Sep 23, 2026
Industries · 6 min read
Vision AI for factory robots, picking from a bin nobody arranged
A cell camera over a bin of oily sheet metal blanks, each found with its position and rotation for the arm, and the new part number labeled before changeover.
Finn Ellingwood · Sep 23, 2026
Industries · 6 min read
Why computer vision pilots fail on the factory floor
A pilot on one inspection station fails for reasons the model never sees: nobody validated it against the inspector, and operators had no way to say no.
Ayman Quadir · Sep 23, 2026
Industries · 6 min read
Visual anomaly detection on the line, catching the defect you never labeled
Learn normal from clean frames, score every fillet against it, route the strange ones to a person, and turn what they name into classes over time.
Rob Hickey · Sep 23, 2026
Operations · 6 min read
What SCADA is and where a camera fits in it
Sensors, PLCs, a network, a server and a historian, and a camera's detection entering the stack as one more tag beside pressure and flow.
Sheikh Srijon · Sep 23, 2026
Case notes · 7 min read
Corrosion detection on transmission insulators, from the inspection pass to a pin you can name
A drone pass along an extra-high-voltage line shows every shed and pin. The frame with early rust looks like the one before it; tracking makes the call stick.
Rob Hickey · Jul 2, 2026
Labeling · 6 min read
Bounding boxes, polygons or masks, pick the label by the question it has to answer
Is there a person at the fence, how much pipe is rusted, where does the field end. Three questions, three label types, and most projects need only the first.
Ayman Quadir · Jun 20, 2026
Product · 6 min read
Auto-labeling from a sentence, and why a person still checks every frame
Type what matters and Lexi drafts a box on every frame. The draft is strong on people and weak on your rust, and the corrections close the gap.
Ayman Quadir · Jun 6, 2026
Edge · 6 min read
Why your best vision model should live at the edge, next to the camera
A substation at the end of a long feeder has a thermal camera, a cellular uplink that drops in rain, and a model that keeps watching when the link does not.
Rob Hickey · May 22, 2026