瑕疵數據收集對於新產品而言極其稀少，一直是重要的議題。然而，瑕疵數據集的品質對深度學習有顯著的影響。本研究致力於應用CNN架構之瑕疵檢測模型，以解決瑕疵資料稀缺性的問題。本研究考量產品生產過程中瑕疵可能會有多種變化，建立能辨識三種不同刮傷瑕疵之瑕疵檢測模型: 一般刮痕、長刮痕與寬刮痕。電腦圖形處理器(顯示卡)外觀瑕疵為驗證本研究衡量模型成效之目標。本研究採用YOLOV3 演算法為架構，所建立之模型經測試可達98.12%平均正確率(mAP)。此外，本研究顯示鋁板適合做為建立GPU外觀刮痕瑕疵資料集之替代材料，透過替代材料，可以降低瑕疵數據收集之成本。本研究藉由迴歸模型找出模型參數(學習率、角度、曝光、飽和)與mAP之關係，且迴歸模式之適配度為97.65%。迴歸模式顯示有一個變數項(角度)與兩個交互作用項(學習率與角度、曝光與飽和)對mAP有顯著性影響。

Defect dataset collection is usually an issue in new products and processes owing to scarcity of defect samples. The quality of defect dataset has significant impact on deep learning performances. This study proposes a convolutional neural network (CNN)-based scratch detection model and deals with the data collection issue of scratch defects. The proposed detection model was designed to identify three types of scratch, namely normal, long and wide scratch, considering the variability of defect shape appearing on the product in a manufacturing process. The graphics processing unit (GPU) is investigated as the target product to evaluate the proposed model. Based on the deep learning algorithm (i.e., You Only Look Once: YOLO), the proposed model is able to reach mean average precision (mAP) of 98.12%. In addition, our experiment result indicates that aluminum is an appropriate alternative material to generate defect data and deal with the scarcity issue of defects data for CNN learning. The data acquisition cost can be significantly reduced by adopting substitutional material to build a variety of scratch defects. This study constructs a relationship between CNN parameters (i.e., learning rate, angle, exposure and saturation) and the resulting mAP by a regression model with high R2 of 97.65%. The regression model shows that mangle effect and the interaction terms (learning rate with angle, exposure with saturation) significantly affect mAP.

Azizah, L. M., Umayah, S. F., Riyadi, S., Damarjati, C., & Utama, N. A. (2017, 24-26 Nov. 2017). Deep learning implementation using convolutional neural network in mangosteen surface defect detection. Paper presented at the 2017 7th IEEE International Conference on Control System, Computing and Engineering (ICCSCE).
Czimmermann, T., Ciuti, G., Milazzo, M., Chiurazzi, M., Roccella, S., Oddo, C. M., & Dario, P. (2020). Visual-based defect detection and classification approaches for industrial applications—a survey. Sensors, 20(5), 1459.
Dai, W., Mujeeb, A., Erdt, M., & Sourin, A. (2020). Soldering defect detection in automatic optical inspection. Advanced Engineering Informatics, 43, 101004. doi:https://doi.org/10.1016/j.aei.2019.101004
Duan, L., Yang, K., & Ruan, L. (2020). Research on Automatic Recognition of Casting Defects Based on Deep Learning. IEEE Access, PP, 1-1. doi:10.1109/ACCESS.2020.3048432
Ebayyeh, A. A. R. M. A., & Mousavi, A. (2020). A Review and Analysis of Automatic Optical Inspection and Quality Monitoring Methods in Electronics Industry. IEEE Access, 8, 183192-183271. doi:10.1109/ACCESS.2020.3029127
Girshick, R. (2015). Fast r-cnn. Paper presented at the Proceedings of the IEEE international conference on computer vision.
Girshick, R., Donahue, J., Darrell, T., & Malik, J. (2014). Rich feature hierarchies for accurate object detection and semantic segmentation. Paper presented at the Proceedings of the IEEE conference on computer vision and pattern recognition.
Hanbay, K., Talu, M. F., & Özgüven, Ö. F. (2016). Fabric defect detection systems and methods—A systematic literature review. Optik, 127(24), 11960-11973. doi:https://doi.org/10.1016/j.ijleo.2016.09.110
Huang, S.-H., & Pan, Y.-C. (2015). Automated visual inspection in the semiconductor industry: A survey. Computers in Industry, 66, 1-10. doi:https://doi.org/10.1016/j.compind.2014.10.006
Hussain, M., Bird, J. J., & Faria, D. R. (2019, 2019//). A Study on CNN Transfer Learning for Image Classification. Paper presented at the Advances in Computational Intelligence Systems, Cham.
Jakubovitz, D., Rodrigues, M. R., & Giryes, R. (2019). Lautum Regularization for Semi-Supervised Transfer Learning. Paper presented at the Proceedings of the IEEE/CVF International Conference on Computer Vision Workshops.
Jordan, M. I., & Mitchell, T. M. (2015). Machine learning: Trends, perspectives, and prospects. Science, 349(6245), 255-260.
LeCun, Y., Bengio, Y., & Hinton, G. (2015). Deep learning. Nature, 521(7553), 436-444. doi:10.1038/nature14539
Liu, B., Zhang, Y., Lv, J., Majeed, A., Chen, C.-H., & Zhang, D. (2021). A cost-effective manufacturing process recognition approach based on deep transfer learning for CPS enabled shop-floor. Robotics and Computer-Integrated Manufacturing, 70, 102128. doi:https://doi.org/10.1016/j.rcim.2021.102128
Lu, J., Behbood, V., Hao, P., Zuo, H., Xue, S., & Zhang, G. (2015). Transfer learning using computational intelligence: A survey. Knowledge-Based Systems, 80, 14-23. doi:https://doi.org/10.1016/j.knosys.2015.01.010
Lu, Y., Zhang, L., & Xie, W. (2020, 22-24 Aug. 2020). YOLO-compact: An Efficient YOLO Network for Single Category Real-time Object Detection. Paper presented at the 2020 Chinese Control And Decision Conference (CCDC).
Ojer, M., Serrano, I., Saiz, F., Barandiaran, I., Gil, I., Aguinaga, D., & Alejandro, D. (2020). Real-time automatic optical system to assist operators in the assembling of electronic components. The International Journal of Advanced Manufacturing Technology, 107(5), 2261-2275. doi:10.1007/s00170-020-05125-z
Rau, H., & Wu, C.-H. (2005). Automatic optical inspection for detecting defects on printed circuit board inner layers. The International Journal of Advanced Manufacturing Technology, 25(9-10), 940-946.
Redmon, J., Divvala, S., Girshick, R., & Farhadi, A. (2016). You only look once: Unified, real-time object detection. Paper presented at the Proceedings of the IEEE conference on computer vision and pattern recognition.
Redmon, J., & Farhadi, A. (2017). YOLO9000: better, faster, stronger. Paper presented at the Proceedings of the IEEE conference on computer vision and pattern recognition.
Ren, S., He, K., Girshick, R., & Sun, J. (2015). Faster r-cnn: Towards real-time object detection with region proposal networks. arXiv preprint arXiv:1506.01497.
Smith, L. N. (2018). A disciplined approach to neural network hyper-parameters: Part 1--learning rate, batch size, momentum, and weight decay. arXiv preprint arXiv:1803.09820.
Tao, F., Zuo, Y., Da Xu, L., & Zhang, L. (2014). IoT-based intelligent perception and access of manufacturing resource toward cloud manufacturing. IEEE Transactions on Industrial Informatics, 10(2), 1547-1557.
Wang, T., Chen, Y., Qiao, M., & Snoussi, H. (2018). A fast and robust convolutional neural network-based defect detection model in product quality control. The International Journal of Advanced Manufacturing Technology, 94(9), 3465-3471.
Ye, R., Chang, M., Pan, C.-S., Chiang, C. A., & Gabayno, J. L. (2018). High-resolution optical inspection system for fast detection and classification of surface defects. International Journal of Optomechatronics, 12(1), 1-10. doi:10.1080/15599612.2018.1444829
Zhang, M., Wu, J., Lin, H., Yuan, P., & Song, Y. (2017). The application of one-class classifier based on CNN in image defect detection. Procedia computer science, 114, 341-348.