YOLOv4-based battery piece defect detection method and system

The invention discloses a battery piece defect detection method based on YOLOv4, and the method comprises the steps: obtaining a battery piece picture with a defect, adding a densely connected three-scale main network in a CSPDarknet53 network of a YOLOv4 model, adding channels for inputting a PANet network, inputting an improved three-scale pyramid structure model into each obtained prediction feature layer, constructing an improved YOLOv4 model, and carrying out the detection of the defect of the battery piece. And inputting a defective battery piece picture into the improved YOLOv4 model to obtain a battery piece defect detection result. The method can improve the accuracy of identifying the defects of the battery piece by the model. 本发明公开了一种基于YOLOv4的电池片缺陷检测方法，包括：获取带有缺陷的电池片图片，在YOLOv4模型的CSPDarknet53网络中，加入密集相连的三尺度主网络，增加输入PANet网络的通道，对获得的每个预测特征层输入改进的三尺度金字塔结构模型，构建改进YOLOv4模型，在改进的YOLOv4模型中输入有缺陷的电池片图片，得到电池片缺陷检测结果。该方法能够提高模型识别电池片缺陷的精确度。