MANUFACTURING
Multi-Sensor Monitoring
Multi-Sensor Monitoring. What it takes to run it.
Fusing thermal and visual data for comprehensive line analysis.
// 01 · The question
“What are the thermal and visual streams saying about the same moment?”
// 02 · Why it is hard
Two sensors, two frame rates, two fields of view. Everything depends on knowing which visual frame belongs with which thermal frame, and that alignment is fragile in ways the images do not show.
// 03 · What the labels have to be
Bounding box
Decided before the first frame
The value is in correspondence, not in outline. A detection in one stream has to be matched to the same physical thing in the other, so boxes plus a registration between the views is what the fusion actually consumes.
This is the decision that is expensive to reverse. The geometry has to match what the answer contains, and finding out it does not means labelling the set a second time.
// 04 · How we run it
Three parts of one loop, on this job.
LexAnnotate
Defect Annotation
Pixel-level defect annotation for surface scratches, cracks, dents, weld faults, and assembly misalignments. Multi-sensor alignment for multi-camera inspection systems.
LexInsight
Line Performance
Connect defect detections to production data - batch numbers, machine IDs, shift schedules. Identify defect patterns across lines and predict maintenance needs before unplanned downtime.
LexAlert
Quality Alerts
Monitor defect detection accuracy across changing production conditions. Detect when lighting shifts, new materials, or equipment aging degrades model performance. Auto-retrain to maintain 99%+ accuracy.
// 05 · What breaks it after launch
This is the use case a pipeline event destroys most completely. A firmware update that shifts a timestamp or re-encodes one stream leaves both feeds looking perfect and the fusion pairing the wrong frames. Rule it out before touching the model.
It is not the only one that can get this use case, it is the one that usually gets it first. All ten conditions.
// 06 · In manufacturing
Defect Detection at Scale
The Challenge
Vision models performed well in pilots but lost accuracy as factory conditions shifted.
Our Execution
Built pixel-level annotated datasets and multi-sensor alignment for reliable defect detection.
The Result
Maintained 99%+ defect detection accuracy across changing production environments. Reduced unplanned downtime.
99%+
Accuracy maintained in production