| abstract |
The present application provides a method and device for defect detection of a circuit board, wherein the method includes: by acquiring a photographed image of the circuit board, and inputting the photographed image into a trained instance segmentation model to obtain a target pixel unit output by the instance segmentation model, the target The pixel unit is a pixel unit that exhibits defects in the captured image; wherein, the instance segmentation model has learned the image features of the defective local circuit board, so as to detect the circuit board defect according to the area ratio of the target pixel unit in the captured image. . Since the instance segmentation model has learned the image features of the defective local circuit board, it can accurately locate the defective target pixel unit in the captured image, and then further calibrate the target pixel unit according to the area ratio of the target pixel unit in the captured image. Test to finally determine whether the circuit board has defects, which helps to improve the accuracy and efficiency of defect detection of the circuit board under test. |