在工業領域，耗材的磨損對產品的品質與生產線的穩定性影響很大，因此對其狀態進行監控非常重要。在刀具加工行業中，砂輪就是一個例子，如果它的表面磨損到一定程度而沒有即時偵測，對生產線的穩定性將是非常有害的。為了能夠對其進行即時監控，最直接的方法就是偵測其表面影像的異常情況。近年來，硬體計算能力已跟上算法，機器學習蓬勃發展。在影像辨識領域，無論是辨識速度或是準確率，都已經遠遠超越人類，並超越了傳統的影像辨識方法。在機器學習領域，需要足夠的標記資料來訓練一個好的模型，但是標記資料的成本是非常昂貴的。為了解決這個問題而產生了半監督學習方法：用少量標記資料訓練一個基本模型，然後用未標記資料標記生成偽標記資料，再用偽標記資料重新訓練模型以提高模型的性能。有很多研究表明這種方法有助於模型訓練。
在本研究中，我們選擇了幾個最先進的深度學習目標偵測模型，並使用半監督學習方法進行模型訓練，建立砂輪表面異常偵測模型，並比較每個模型的性能。實驗結果表明，在此情境中，使用YOLOv4-CSP模型所建立的砂輪表面異常偵測模型的性能最好。另一方面，使用半監督學習方法有助於提高某些模型的性能。

In the tool processing industry, if the grinding wheel's surface wears to a certain level and it is not detected in real-time, it will be very harmful to the stability of the production line. In order to be able to monitor in real-time, the most direct way is to detect anomalies on its surface. In recent years, hardware computing power has kept up with algorithms, and machine learning has flourished. In the domain of image recognition, regardless of the recognition speed and accuracy, it has far surpassed humans and surpassed traditional image recognition. In the field of machine learning, sufficient labeled data is needed to train a good model, but the cost of labeling the data is very expensive. To solve this problem, a semi-supervised learning method was developed in this thesis to build the model with few labeled data. Many studies that have shown this method is helpful to model training.
In this thesis, we select several state-of-the-art deep learning object detection models, and use semi-supervised learning methods for model training. Experimental results show that YOLOv4-CSP model has the best performance and semi-supervised learning methods is indeed helpful to improve performance in some models.

N. Arunachalam and B. Ramamoorthy. Texture analysis for grinding wheel wear assessment using machine vision. Proceedings of the Institution of Mechanical Engineers, Part B: Journal of Engineering Manufacture, 221(3):419–430, 2007.
A. Bochkovskiy, C. Wang, and H. M. Liao. Yolov4: Optimal speed and accuracy of object detection. CoRR, abs/2004.10934, 2020.
E. Brinksmeier and F. Werner. Monitoring of grinding wheel wear. CIRP annals, 41 (1):373–376, 1992.
O. Chapelle, B. Scholkopf, and A. Zien. Semi-supervised learning (chapelle, o. et al., eds.; 2006)[book reviews]. IEEE Transactions on Neural Networks, 20(3):542–542, 2009.
R. Girshick, J. Donahue, T. Darrell, and J. Malik. Rich feature hierarchies for accurate object detection and semantic segmentation. 2014.
F. Guo, Y. Qian, and Y. Shi. Real-time railroad track components inspection based on the improved yolov4 framework. Automation in Construction, 125:103596, 2021.
L. Guo, N. Li, F. Jia, Y. Lei, and J. Lin. A recurrent neural network based health indicator for remaining useful life prediction of bearings. Neurocomputing, 240:98–109, 2017.
K. Han, M. Sun, X. Zhou, G. Zhang, H. Dang, and Z. Liu. A new method in wheel hub surface defect detection: Object detection algorithm based on deep learning. 2017.
K. He, X. Zhang, S. Ren, and J. Sun. Spatial pyramid pooling in deep convolutional networks for visual recognition. CoRR, abs/1406.4729, 2014.
K. He, X. Zhang, S. Ren, and J. Sun. Spatial pyramid pooling in deep convolutional networks for visual recognition. IEEE transactions on pattern analysis and machine intelligence, 37(9):1904–1916, 2015.
K. He, X. Zhang, S. Ren, and J. Sun. Deep residual learning for image recognition. 2016.
G. Hinton, T. Sejnowski, and H. Sejnowski. Unsupervised Learning: Foundations of Neural Computation. A Bradford Book. Bradford University Press, 1999.
T. Hui, Y. Xu, and R. Jarhinbek. Detail texture detection based on yolov4-tiny combined with attention mechanism and bicubic interpolation. IET Image Processing, 2021.
S. Ioffe and C. Szegedy. Batch normalization: Accelerating deep network training by reducing internal covariate shift. arXiv preprint arXiv:1502.03167, 2015.
K. Kannan, N. Arunachalam, A. Chawla, and S. Natarajan. Multi-sensor data analytics for grinding wheel redress life estimation-an approach towards industry 4.0. Procedia Manufacturing, 26:1230–1241, 2018.
Y. LeCun, P. Haffner, L. Bottou, and Y. Bengio. Object recognition with gradient-based learning. In D. Forsyth, J. Mundy, V. di Gesu, and R. Cipolla, editors, Shape, Contour and Grouping in Computer Vision, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), pages 319–345. Springer Verlag, 1999.
D.-H. Lee et al. Pseudo-label: The simple and efficient semi-supervised learning method for deep neural networks. 2013.
Y. Lei, N. Li, L. Guo, N. Li, T. Yan, and J. Lin. Machinery health prognostics: A systematic review from data acquisition to rul prediction. Mechanical Systems and Signal Processing, 104:799–834, 2018.
P. Lezanski. A data-driven predictive model of the grinding wheel wear using the neural network approach. Journal of Machine Engineering, 17, 2017.
Y. Li, H. Huang, Q. Xie, L. Yao, and Q. Chen. Research on a surface defect detection algorithm based on mobilenet-ssd. Applied Sciences, 8(9):1678, 2018.
S. Liu, L. Qi, H. Qin, J. Shi, and J. Jia. Path aggregation network for instance segmentation, 2018.
J. Redmon and A. Farhadi. Yolov3: An incremental improvement. arXiv preprint arXiv:1804.02767, 2018.
J. Redmon, S. Divvala, R. Girshick, and A. Farhadi. You only look once: Unified, real-time object detection. 2016.
S. Ren, K. He, R. Girshick, and J. Sun. Faster r-cnn: Towards real-time object detection with region proposal networks. arXiv preprint arXiv:1506.01497, 2015.
S. Russell and P. Norvig. Artificial Intelligence: A Modern Approach. Prentice Hall, 3 edition, 2010.
J. Su and Y. Tarng. Measuring wear of the grinding wheel using machine vision. The International Journal of Advanced Manufacturing Technology, 31(1-2):50–60, 2006.
D. Tabernik, S. ˇ Sela, J. Skvarˇc, and D. Skoˇcaj. Segmentation-based deep-learning approach for surface-defect detection. Journal of Intelligent Manufacturing, 31(3): 759–776, 2020.
M. Tan and Q. V. Le. Efficientnet: Rethinking model scaling for convolutional neural networks. CoRR, abs/1905.11946, 2019.
X. Tao, D. Zhang, W. Ma, X. Liu, and D. Xu. Automatic metallic surface defect detection and recognition with convolutional neural networks. Applied Sciences, 8(9): 1575, 2018.
A.-A. Tulbure, A.-A. Tulbure, and E.-H. Dulf. A review on modern defect detection models using dcnns–deep convolutional neural networks. Journal of Advanced Research, 2021.
C. Wang, H. M. Liao, I. Yeh, Y. Wu, P. Chen, and J. Hsieh. Cspnet: A new backbone that can enhance learning capability of CNN. CoRR, abs/1911.11929, 2019.
C. Wang, A. Bochkovskiy, and H. M. Liao. Scaled-yolov4: Scaling cross stage partial network. CoRR, abs/2011.08036, 2020a.
S. Wang, X. Xia, L. Ye, and B. Yang. Steel surface defect detection using transfer learning and image segmentation. 2020b.
S. Yanan, Z. Hui, L. Li, and Z. Hang. Rail surface defect detection method based on yolov3 deep learning networks. 2018.
Z. Zheng, P. Wang, W. Liu, J. Li, R. Ye, and D. Ren. Distance-iou loss: Faster and better learning for bounding box regression. CoRR, abs/1911.08287, 2020.