This project is the client's first AI appearance multi-defect inspection project that successfully passed their ARR test.
The client is a major wireless telecom supplier in China. The appearance inspection of the production line is complicated as there are many types of defects to detect: crescent deviation, cracks, wrinkles, foreign objects, offset QR code, missing QR code, poorly printed QR code, etc. It is time-consuming and requires a lot of manpower to conduct visual inspections, which directly affects overall production efficiency. The client expects to use automation equipment to replace labor.
Achieve full AI appearance defect detection for wireless charging coil on this product line, pass rate over 99%
Solutions and Results
SMore ViMo used classification, segmentation, QR code recognition, and other modules to create AI appearance defect detection automation equipment. The equipment provides OK/NG material differentiation, 14 types of small defect detection for NG materials, and 6 types of sub-assembly trays, comprehensively improving the production line inspection efficiency while providing analysis data of the manufacturing process.
● Pass rate ≥ 99.0%
● Overkill rate ≤ 1%
● Processing speed ≤ 0.3 seconds/piece; process 100,000+ pieces daily on average
● Achieve AI full inspection; inspection efficiency increased by 2 times