人工智慧之深度學習技術已逐步應用於生產線上之產品瑕疵檢測，深度學習需使用大量影像訓練才有良好的辨識效果，然而在實際的產線中，瑕疵品相對良品的比例低很多，造成瑕疵品的訓練樣本數過少，導致深度學習的檢測效果不理想，如能克服此問題，將可大幅提升將深度學習應用於產線的可能性。本研究針對樣本數量不多的鍍鎳金屬片，提出利用旋轉法、分割法與增量標註法進行數據集擴增，然後藉由YOLO模型進行訓練與辨識，經過辨識後，影像辨識率達到87.5 %，接著搭配拼圖法影像辨識流程，將影像辨識率進一步提升至97.38 %。
鍍鎳金屬片之瑕疵檢測系統實際檢測結果顯示，良品之誤檢率為16.67 %、瑕疵品之漏檢率僅為1.54 %，而檢測準確率可達到95.62 %，證明本系統實際應用於檢測金屬片已有良好的檢測效果。

The technology of deep learning in artificial intelligence has been gradually applied to the inspection of products’ defects on a production line. Deep learning requires a lot of image data training to have a good recognition effect. However, in an actual production line, the quantity of defective products is much lower than that of the good products. Therefore the number of training samples is far from needed, which leads to the unsatisfactory inspection effect of deep learning. If this problem can be solved, the possibility of applying deep learning to a production line will be greatly improved. In this thesis, for the nickel-plated metal sheet with a small number of samples, it is proposed to use the rotation method, the segmentation method and the incremental labeling method to expand the data set, and then use the YOLO model for training and identification. After the identification, the image recognition rate reaches 87.5 %, and then use the puzzle image recognition process to further increase the image recognition rate to 97.38%.
The actual test results of the defect inspection system for nickel-plated metal sheets show that the false inspection rate of good products is 16.67%, the missed inspection rate of defective products is only 1.54%, and the inspection accuracy rate can reach 95.62%, which proves that the system has good effects on inspection of defects on the surface of nickel-plated metals.

[1] Cai, N., Cen, G., Wu, J., Li, F., Wang, H. and Chen, X. (2018), “SMT solder joint inspection via a novel cascaded convolutional neural network,” IEEE Transactions on Components, Packaging and Manufacturing Technology, Vol. 8, No. 4, pp. 670-677.
[2] Lin, H., Li, B., Wang, X., Shu, Y. and Niu, S. (2019), “Automated defect inspection of LED chip using deep convolutional neural network,” Journal of Intelligent Manufacturing, Vol. 30, No. 6, pp. 2525-2534.
[3] Taylor, L. and Nitschke, G. (2018), “Improving deep learning with generic data augmentation,” IEEE Symposium Series on Computational Intelligence, Nov. 18-21, 2018, Bengaluru, India, pp. 1542-1547.
[4] http://www.vision.caltech.edu/Image_Datasets/Caltech101/
[5] Krizhevsky, A., Sutskever, I. and Hinton, G. E. (2012), “Imagenet classification with deep convolutional neural networks,” Advances in neural information processing systems, Vol. 25, pp. 1097-1105.
[6] DeVries, T. and Taylor, G. W. (2017), “Improved regularization of convolutional neural networks with cutout,” arXiv:1708.04552.
[7] Zhang, H., Cisse, M., Dauphin, Y. N. and Lopez-Paz, D. (2017), “mixup: Beyond empirical risk minimization,” arXiv:1710.09412.
[8] Yun, S., Han, D., Oh, S. J., Chun, S., Choe, J. and Yoo, Y. (2019), “Cutmix: Regularization strategy to train strong classifiers with localizable features,” International Conference on Computer Vision, Oct. 27-Nov. 2, 2019, Seoul, Korea, pp. 6023-6032.
[9] Bochkovskiy, A., Wang, C. Y. and Liao, H. Y. M. (2020), “Yolov4: Optimal speed and accuracy of object detection,” arXiv:2004.10934.
[10] Zhu, X., Liu, Y., Qin, Z. and Li, J. (2017), “Data augmentation in emotion classification using generative adversarial networks,” arXiv:1711.00648.
[11] Goodfellow, I. J., Pouget-Abadie, J., Mirza, M., Xu, B., Warde-Farley, D., Ozair, S., Courville, A. and Bengio, Y. (2014), “Generative adversarial nets,” International Conference on Neural Information Processing Systems, Dec. 8-13, 2014, Montreal, Canada, pp. 2672-2680.
[12] Zhu, J. Y., Park, T., Isola, P. and Efros, A. A. (2017), “Unpaired image-to-image translation using cycle-consistent adversarial networks,” International Conference on Computer Vision, Oct. 27-29, 2017, Venice, Italy, pp. 2223-2232.
[13] Cubuk, E. D., Zoph, B., Mané, D., Vasudevan, V. and Le, Q. V. (2019), “AutoAugment: Learning Augmentation Strategies From Data,” IEEE Conference on Computer Vision and Pattern Recognition, June 16-20, Long Beach, CA, USA, pp. 113-123.
[14] Zoph, B. and Le, Q. V. (2016), “Neural architecture search with reinforcement learning,” arXiv:1611.01578
[15] Redmon, J., Divvala, S., Girshick, R. and Farhadi, A. (2016), “You Only Look Once: Unified, real-time object detection,” IEEE Conference on Computer Vision and Pattern Recognition, June 27-30, 2016, Las Vegas, NV, USA, pp. 779-788.
[16] He, Y., Song, K., Meng, Q. and Yan, Y. (2019), “An end-to-end steel surface defect detection approach via fusing multiple hierarchical features,” IEEE Transactions on Instrumentation and Measurement, Vol. 69, No. 4, pp. 1493-1504.
[17] Song, L., Li, X., Yang, Y., Zhu, X., Guo, Q. and Yang, H. (2018), “Detection of micro-defects on metal screw surfaces based on deep convolutional neural networks,” Sensors, Vol. 18, pp. 1-14.
[18] Lecun, Y., Bottou, L., Bengio, Y. and Haffner, P. (1998), ”Gradient- based learning applied to document recognition,” Proceedings of the IEEE, Vol. 86, No. 11, pp. 2278-2324.
[19] Sun, X., Gu, J., Huang, R., Zou, R. and Giron, P.,B. (2019), “Surface defects recognition of wheel hub based on improved faster R-CNN,” Electronics, Vol. 8, No. 5, pp. 481-497.
[20] S. Ren, K. He, R. Girshick and J. Sun(2017), “Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks,” Institute of Electrical and Electronics Engineers Transactions on Pattern Analysis and Machine Intelligence, Vol. 39, No. 6, pp. 1137-1149.
[21] Li, Y., Huang, H., Xie, Q., Yao, L. and Chen, Q. (2018), “Research on a surface defect detection algorithm based on MobileNet-SSD,” Applied Sciences, Vol. 8, No. 9, pp. 1678.-1695.
[22] Liu, W., Anguelov, D., Erhan, D., Szegedy, C., Reed, S., Fu, C-Y. and Berg, A.C. (2016), “SSD: Single Shot MultiBox Detector,” European conference on computer vision, Oct. 11-14, 2016, Amsterdam, The Netherlands, pp. 21-37.
[23] Li, J., Su, Z., Geng, J. and Yin, Y. (2018), “Real-time detection of steel strip surface defects based on improved yolo detection network,” IFAC-PapersOnLine, Vol. 51, No. 21, pp. 76-81.
[24] Chen, J., Liu, Z., Wang, H., Núñez, A. and Han, Z. (2017), “Automatic defect detection of fasteners on the catenary support device using deep convolutional neural network,” IEEE Transactions on Instrumentation and Measurement, Vol. 67, No. 2, pp. 257-269.
[25] Bochkovskiy, A., Wang, C. Y. and Liao, H. Y. M. (2021), “Scaled-yolov4: Scaling cross stage partial network,” IEEE Conference on Computer Vision and Pattern Recognition, June 19-25, pp. 13029-13038.