Our self developed deep learning network model,the mAR index of our model is 0.99 ~ 1.0, which is 0.9 relative to the industry average value of target detection
【Machine Principle】 Image acquisition system cooperate with mechanical movement to collect samples, and machine summarize defect features by machine learning. After a large number of learning and correction, the equipment can reach the level far beyond human visual inspection. According to different products, escape rate tends to 0 and escape defect means extreme slight defect(extreme slight defect means most of the experienced inspector can not find defect base on fixed duration, lux, distance, uninformed testing, defect have no influence on function)
