Visual detection and identification device for metal workpieces
By using a vertically arranged track mechanism and a deep neural network to adjust the camera and light source angles, the problem of poor adaptability of traditional visual inspection equipment is solved, enabling efficient and accurate inspection of metal workpieces, especially the identification of weakly significant defects.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SHENZHEN JPR TECH CO CTD
- Filing Date
- 2026-04-08
- Publication Date
- 2026-07-21
AI Technical Summary
Traditional visual inspection equipment is difficult to adapt to the inspection needs of various metal workpiece types and defects, and lacks intelligent adjustment mechanisms, which leads to the determination of inspection parameters relying on human experience and low detection rate of weakly significant defects.
The system employs a vertically arranged first and second arc-shaped track mechanism, combined with a deep neural network angle decision model, to automatically adjust the camera and light source angles. It matches the optimal detection parameters through a storage module and uses a multi-angle fusion imaging strategy to identify defects.
It enables rapid adaptation to multi-variety, small-batch testing, improves the flexibility and accuracy of the testing system, significantly increases the detection rate of weakly significant defects, and reduces the need for manual adjustments.
Smart Images

Figure CN122430239A_ABST