Manufacturing & automotive
Visual AI on the line, with a person behind every hard call.
Defect and deviation annotation, and live review of production line models around the clock, for factories, automotive and EV plants.
From the production line to the quality report
Every part of the workflow where we can take work off your team.
Inspection
Process
Assets
Safety
We work with line video, images and thermal.
What we label
Every class is written into a guide with your team before the first batch. Files come back as COCO JSON, YOLO, Pascal VOC, CSV, JSONL or your own schema.
Line images and video
Defects
Polygon and bounding box
Scratch, dent, crack, missing part, contamination
Assembly
Class and keypoints
Correct, misaligned, missing component
Packaging
Bounding box and OCR
Label present, print error, seal fault, batch code
Equipment
Gauges and meters
OCR and keypoints
Needle position, digital reading
Condition
Polygon and severity
Leak, rust, worn belt
Alerts
Deviations
Class and reason
Real deviation, false positive, needs operator
How we test the model
The tests that matter for manufacturing & automotive models, and what you get back from each.
False positive rate
Every reviewed deviation recorded as real or false.
You get backFalse positives by line, shift and defect class
Rare defect recall
Rare defects collected and reviewed as their own set.
You get backRecall on rare classes
Quality against target
Reviewer accuracy measured against the target you set.
You get backQuality by week
Where manufacturing & automotive teams start
Problems teams bring us, and the first piece of work we would scope.
The line cannot stop every time the model is unsure.
What we doReviewers decide on flagged deviations in under two minutes, 24/7, as we do for Client E.Live reviewGood parts outnumber defects by thousands to one.
What we doRare defects collected and labelled on purpose, with a class guide your quality team signs off.AnnotateDefect definitions differ between plants.
What we doOne guide across sites, with disputed examples settled with your quality engineers.AnnotateRelated work
Human in the loop support for visual deviation detection on production lines, with a response time requirement under two minutes.
24/7 live review for production line visual AI, quality above 95% by month two
Twelve times the volume in five weeks, with quality holding at 95 to 98%
2,000 robot videos described every week by a team built in seven days
Half a million images for autonomous navigation, with every file kept on EU servers
Bring us the manufacturing & automotive data your model gets wrong.
Send a small batch and get it back done, free, usually within 24 to 48 hours. You judge the work before you pay for anything.