During the reflow process, excessive solder paste or printing offset may cause solder balls to short circuit. In the past, such defects were mostly detected by manual visual inspection, which was inefficient and affected the production efficiency. Since the flow pattern of excess solder paste under high temperature is unpredictable, it is also difficult to detect by traditional optical inspection AOI.Using the Instance Segmentation technology of Solomon SolVision AI image platform, the reflow short circuit defects in the image samples are located and annotated, and then used to train the AI model. The trained model can easily detect the short circuit between adjacent solder balls.