47 / Hardware & IoT
AI X-Ray Defect Inspection for Metal Castings
We built an X-ray inspector for metal castings that spots defects right on the production line, running on edge hardware so the X-ray images never have to leave the plant.
- Built by
- 360 Labs (Internal)
- Stack & approach
- Computer VisionX-RayDefect DetectionEdge InferenceOn-Prem
The problem
Defect inspection of castings depends on operator judgement, and image data cannot leave the plant for cloud processing.
Solution & architecture
An on-device vision model inspects X-ray captures at the line, flagging defects in real time with no cloud round-trip. The model is trained on plant data and deployed on edge hardware, so defect flags route into line QC with no image egress from the facility.
Why it is built this way
Inference on the line
Edge hardware, no cloud round-trip
Images never leave
No X-ray egress from the plant
Trained on plant data
Tuned to the real production line
