在顯卡組裝出貨的製程中，在散熱鰭片與PCB板的貼合面貼上散熱貼片幫助顯卡散熱是重要且必要的一個環節，然由於整個過程是由工人逐張檢查，不免還是會有所缺漏。因此本論文基於YOLOv7設計了一個散熱貼片檢測的系統雛形，並以此設計了兩個模組以達成動態檢測散熱貼片的目標。顯卡追蹤模組透過背景相減法判斷相機視野內是否有物體經過並透過網路傳輸影像至伺服器端。在伺服器端則是透過散熱貼片檢測模組檢測影像中是否包含有膜散熱貼片、無膜以及殘膠，在檢測結束後回傳結果至邊緣設備，邊緣設備透過檢測前於使用者介面的設定作為判斷的依據顯示結果。在YOLOv7的訓練的部分則是透過分析餘本論文設計的資料集在透過影像處理的方式進行旋轉下，什麼角度可使模型有較好的訓練效果。在考量到相機以及輸送帶速度若產生不匹配的問題時，會產生動態模糊的影像，因此在資料集中加入的模糊影像，確認於使用這種資料集訓練出的模型各種模糊度以及模糊角度的判斷能力。

In the manufacturing process of graphics card assembly and shipment, it is important and necessary to apply thermal patches to the thermal fins and PCB boards to help dissipate the heat of the graphics card, however, since the whole process is checked by workers one by one, it is inevitable that there will still be some omissions. Therefore, based on YOLOv7, this paper designs a prototype of the thermal patch detection system, and two modules are designed to achieve the goal of dynamic detection of thermal patches. The video card tracking module determines whether there is an object passing in the camera's field of view by background subtraction and transmits the image to the server over the network. On the server side, the heat sink detection module detects whether the image contains a film heat sink, no film, or glue residue, and then sends the results back to the edge device, which displays the results based on the user-interface settings prior to the detection. In the training part of YOLOv7, we analyzed the data set designed in the rest of this thesis by analyzing what angle can make the model have a better training effect under the rotation through the image processing method. Considering that a mismatch between the camera and the conveyor belt speed will produce dynamic blurred images, the blurred images added to the dataset confirm the ability of the model trained with this dataset to judge various blurring degrees and blurring angles.

[1] Y. Joon Lee, "Thermo-mechanical properties of high performance thermal interface gap filler pads," 2010 12th IEEE Intersociety Conference on Thermal and Thermomechanical Phenomena in Electronic Systems, Las Vegas, NV, USA, 2010, pp. 1-8, doi: 10.1109/ITHERM.2010.5501306.
[2] Nakayama, W., and Bergles, A. E., "Thermal Interfacing Techniques for Electronic Equipment—A Perspective ." ASME. J. Electron. Packag. June 2003; 125(2): 192–199.
[3] R. Prasher, "Thermal Interface Materials: Historical Perspective, Status, and Future Directions," in Proceedings of the IEEE, vol. 94, no. 8, pp. 1571-1586, Aug. 2006, doi: 10.1109/JPROC.2006.879796.
[4] J. Redmon, S. Divvala, R. Girshick and A. Farhadi, "You Only Look Once: Unified, Real-Time Object Detection," 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Las Vegas, NV, USA, 2016, pp. 779-788, doi: 10.1109/CVPR.2016.91.
[5] Chien-Yao Wang, Alexey Bochkovskiy, and Hong-Yuan Mark Liao. "YOLOv7: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors." arXiv:2207.02696, 2022.
[6]Ningning Ma, Xiangyu Zhang, Hai-Tao Zheng, Jian Sun."ShuffleNet V2: Practical Guidelines for Efficient CNN Architecture Design" arXiv:1807.11164,2018.
[7] X. Ding, Y. Guo, G. Ding and J. Han, "ACNet: Strengthening the Kernel Skeletons for Powerful CNN via Asymmetric Convolution Blocks," 2019 IEEE/CVF International Conference on Computer Vision (ICCV), Seoul, Korea (South), 2019, pp. 1911-1920, doi: 10.1109/ICCV.2019.00200.
[8] Shuxuan Guo, Jose M. Alvarez, Mathieu Salzmann. "ExpandNets: Linear Over-parameterization to Train Compact Convolutional Networks"arXiv:1811.10495,2018.
[9] N. Nagabandi et al., "Metallic nanocomposites as next-generation thermal interface materials," 2017 16th IEEE Intersociety Conference on Thermal and Thermomechanical Phenomena in Electronic Systems (ITherm), Orlando, FL, USA, 2017, pp. 400-406, doi: 10.1109/ITHERM.2017.7992502.
[10] Kaiming He, Xiangyu Zhang, Shaoqing Ren, Jian Sun."Deep Residual Learning for Image Recognition" arXiv:1512.03385,2015.
[11] Zheng Ge, Songtao Liu, Feng Wang, Zeming Li, Jian Sun. "YOLOX: Exceeding YOLO Series in 2021" arXiv:2107.08430,2021.
[12] https://docs.opencv.org/3.4/d4/dee/tutorial_optical_flow.html,2023/07/25.
[13] Guo-Wu Yuan, Jian Gong, Mei-Ni Deng, Hao Zhou and Dan Xu,"A Moving Objects Detection Algorithm Based on Three-Frame Difference and Sparse Optical Flow" Information Technology Journal, 2014 ;13 , pp.1863-1867.
[14] C. Kim, J. Cho and Y. Lee, "The Relational Properties among Results of Background Subtraction," 2008 10th International Conference on Advanced Communication Technology, Gangwon, Korea (South), 2008, pp. 1887-1890, doi: 10.1109/ICACT.2008.4494153.
[15] Barron, J.L., Fleet, D.J. & Beauchemin, S.S.,"Performance of optical flow techniques. ", Int J Comput Vision 12, 1994, pp. 43–77, doi: 10.1007/BF01420984
[16] John H. Holland, "Adaptation in Natural and Artificial Systems: An Introductory Analysis with Applications to Biology, Control and Artificial Intelligence", MIT Press, Cambridge, MA, USA, 1992.
[17] Chi-Hsu Wang, Han-Leih Liu and Chin-Teng Lin, "Dynamic optimal learning rates of a certain class of fuzzy neural networks and its applications with genetic algorithm," in IEEE Transactions on Systems, Man, and Cybernetics, Part B (Cybernetics), vol. 31, no. 3, pp. 467-475, June 2001, doi: 10.1109/3477.931548.
[18] N. Gorgolis, I. Hatzilygeroudis, Z. Istenes and L. –. G. Gyenne, "Hyperparameter Optimization of LSTM Network Models through Genetic Algorithm," 2019 10th International Conference on Information, Intelligence, Systems and Applications (IISA), Patras, Greece, 2019, pp. 1-4, doi: 10.1109/IISA.2019.8900675.
[19] Praveen Ranjan Srivastava and Tai-hoon Kim, “Application ofgenetic algorithm in
software testing” International Journal ofsoftware Engineering and its Applications, 3(4),
2009, pp.87- 96.
[20] G. A. Vignaux and Z. Michalewicz, "A genetic algorithm for the linear transportation problem," in IEEE Transactions on Systems, Man, and Cybernetics, vol. 21, no. 2, pp. 445-452, March-April 1991, doi: 10.1109/21.87092.
[21] Chien-Yao Wang, Hong-Yuan Mark Liao, I-Hau Yeh, "Designing Network Design Strategies Through Gradient Path Analysis", arXiv:2211.04800,2022.
[22] S. Lessmann, R. Stahlbock, and S. F. Crone, "Optimizing hyperparameters of support vector machines by genetic algorithms," in IC-AI, 2005, pp. 74-82.
[23] A. S. Wicaksono and A. A. Supianto, "Hyper Parameter Optimization using Genetic Algorithm on Machine Learning Methods for Online News Popularity Prediction,"vol. 9, no. 12, 2018, pp. 263-267.
[24] Alex Krizhevsky, Ilya Sutskever, and Geoffrey E. Hinton. 2017. ImageNet classification with deep convolutional neural networks. Commun. ACM 60, 6 (June 2017), 84–90, doi:
10.1145/3065386