Manufacturing · Edge · Case Study

VLM-based defect detection on the line - replaced a legacy CV system on Jetson

A manufacturer was running a 2018-era CV defect detector that needed retraining for every new product variant. We replaced it with a fine-tuned Qwen-VL deployed on Jetson Orin at every inspection station - generalising across variants without retraining.

94%
defect recall (legacy CV: 71%)
180ms
per-frame inference on Jetson
0
per-variant retraining cycles
62%
drop in warranty claims attributable to missed defects
Problem

What they were stuck on

The client's existing defect-detection system was a classical CNN that needed weeks of relabelling and retraining for every new product variant. False-negative rate was 29% on edge cases. They were losing on warranty claims and customer trust. They wanted a system that could understand “what counts as a defect” the way a human inspector would - and ship in a form factor that bolts onto the line.

Approach

How we built it

STEP 01

Labelled defect corpus

Built a multimodal dataset: image + text description of the defect, with severity tags. Pulled from 6 years of QA archives plus 12 weeks of fresh capture. ~40k labelled examples across 200 defect classes.

STEP 02

Qwen-VL fine-tune

QLoRA fine-tune of Qwen-VL-7B on the defect corpus. Trained the model to output structured JSON: { has_defect, type, severity, location, confidence }.

STEP 03

Edge quantization

INT4 AWQ quantization plus TensorRT compilation for the vision encoder. Fits in 12GB VRAM with headroom for the image-processing pipeline.

STEP 04

Jetson deployment

Each inspection station runs a Jetson Orin AGX with the model, a camera trigger, and a custom runtime. Sub-180ms per frame at 1080p. No internet required.

STEP 05

Continuous learning loop

Edge devices send only their low-confidence cases (with PII-stripped frames) to a central retraining queue. Models redeploy OTA every 2 weeks with new variant coverage - no per-variant retraining cycle.

Stack

What we used

QwenQwen-VL 7BQLoRAAWQ INT4 quantizationNvidiaTensorRTJetson Orin AGXOTA model rollout
Outcomes

What changed

94%defect recall (legacy CV: 71%)
180msper-frame inference on Jetson
0per-variant retraining cycles
62%drop in warranty claims attributable to missed defects

We used to retrain for every new product. Now the model reads the defect description like an inspector would - and we ship new variants on day one.

- Plant manager, industrial manufacturer (name withheld)

Have a similar problem? Let's scope it.

A 30-minute call. We'll tell you whether we can help - and if not, who can.

Talk to us