Realizing real-time printed circuit board(PCB) surface defect detection is the basis for improving the intelligence of the PCB fabrication process. Aiming at the original PCB inspection method which is time-consuming, low-accuracy and not easy to move, this paper proposes an improved model based on YOLOv7-tiny. Replace the SiLU activation function in YOLOv7-tiny with the ELU function; introduce a centralized integrated convolutional module (C3 module), and combine depthwise separable convolution with C3 to form a centralized integrated depthwise separable convolution module; and add a convolutional block attention module. Experimentally, the improved model performs well in detection accuracy, recall rate, and mean average precision, and the size of the model drops by 2.8 MB compared to the original model. It also shows better detection results when compared with other mainstream target detection schemes. The ability of the improved YOLOv7-tiny to maintain higher accuracy while also reducing the memory requirements of the model opens up new possibilities for real-time detection of PCB defects as well as edge deployment.
LIKui, CHENManlong, YANLizhi, et al. Surface quality detection method of graphite seal ring based on machine vision[J]. Journal of Shaanxi University of Technology (Natural Science Edition), 2021, 37(2): 29-34. (in Chinese)
[7]
JUDIH M, KAPOYAHK, ASHAARIN S. Visualising automatic product inspection of PCB units[C]// 2009 International Conference on Electrical Engineering and Informatics, 2009: 204-207.
LUODongliang, CAIYuxuan, YANGZihao, et al. Survey on industrial defect detection with deep learning[J]. Scientia Sinica(Informationis), 2022, 52(6): 1002-1039. (in Chinese)
[10]
HINTONG, DENGL, YUD, et al. Deep neural networks for acoustic modeling in speech recognition: the shared views of four research groups[J]. IEEE Signal Processing Magazine, 2012, 29(6): 82-97.
WANGHe, FANXing. A lightweight object detection algorithm based on improved SSD[J]. Journal of Test and Measurement Technology, 2023, 37(2): 152-157. (in Chinese)
LUORenze, TANGXiang, YUHong, et al. Weld defect detection method of ray image based on improved Faster RCNN[J]. Electronic Measurement Technology, 2023, 46(22): 160-168. (in Chinese)
[15]
DUJ, LUH, HUM, et al. CNN‐based infrared dim small target detection algorithm using target‐oriented shallow‐deep features and effective small anchor[J]. IET Image Processing, 2021, 15(1): 1-15.
LIUFang, LIUYukun, LINSen, et al. Fast recognition method for tomatoes under complex environments based on improved YOLO[J]. Transactions of the Chinese Society for Agricultural Machinery, 2020, 51(6): 229-237. (in Chinese)
ZHAOChunjiang, LIANGXuewen, YUHelong, et al. Automatic identification and counting method of caged hens and eggs based on improved YOLO v7[J]. Transactions of the Chinese Society for Agricultural Machinery, 2023, 54(7): 300-312. (in Chinese)
ZHANGLifeng, TIANYing. Improved YOLOv8 multi-scale and lightweight vehicle object detection algorithm[J]. Computer Engineering and Applications, 2024, 60(3): 129-137. (in Chinese)
[26]
WANGC, YEHI, LIAOH. Yolov9: learning what you want to learn using programmable gradient information[C]//IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2024.
[27]
PARKH, YOOY, SEOG, et al . C 3: concentrated-comprehensive convolution and its application to semantic segmentation[C]//IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2018.
[28]
CHOLLET, F. Xception: deep learning with depthwise separable convolutions[C]//2017 IEEE Conference on Computer Vision and Pattern Recognition, 2017: 1800-1807.
[29]
WOO S, PARKJ, LEEJ Y, et al. CBAM: convolutional block attention module[C]//European Conference on Computer Vision, 2018: 3-19.