Visual inspection
Defect detection solutions and cases
Glass defect detection
Module glass defect detection-POL sticker
CELLCUT defect detection
LCD panel defect detection software process
Algorithm advantages
After years of intensive cultivation, we have an abnormal database for each process segment. The database contains various forms of defects such as chipping, cracks, dirt, and water droplets, as well as data that are prone to false alarms such as foreign matter, photoresist residues, marks, and process residues. We use AI deep learning classifiers to classify defects and interference items, with high classification accuracy, ensuring a high detection rate and a low false detection rate;
Adopt the evolutionary learning method (database update -> AI evolutionary learning -> evolutionary model) to deal with missed detections and false detections, which greatly improves the efficiency of exception handling and simplifies the exception handling process;
Traditional algorithms perform initial defect location, and deep learning algorithms perform fine classification of defects while ensuring processing efficiency and detection rate;
Using the deep learning algorithm library to perform int8 quantization acceleration on the model, the speed is increased by nearly 4 times, ensuring the real-time performance of the system;