Method for realizing BPCB surface defect detection based on CNN

• Main Authors: YANG HAIDONG, ZHANG PENGZHONG
• Format: Patent
• Language: Chinese; English
• Published: 29.11.2022
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; IMAGE DATA PROCESSING OR GENERATION, IN GENERAL; INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIRCHEMICAL OR PHYSICAL PROPERTIES; MEASURING; PHYSICS; TESTING

The invention belongs to the field of artificial intelligence technology application, and particularly relates to a method for realizing BPCB surface defect detection based on CNN, and the method comprises the steps: obtaining a training picture data set containing BPCB surface normal picture samples and BPCB surface picture defect samples; preprocessing the training picture data set to obtain a pre-training image data set; training a preset multi-target detection network according to the pre-training data set, and outputting and storing the trained preset multi-target detection network; and performing surface defect detection on the BPCB picture by using the preset multi-target detection network. According to the method, the neural network structure obtained by integrating the feature extraction network, the detector and the classification network device is combined, and defect detection and classification tasks in the defect detection process of the BPCB can be separated, so that the detection accuracy is i