在現代電子產品中，印刷電路板是核心組件，其瑕疵檢測對於產品品質與可靠性至關重要。然而，傳統檢測方式，如人工檢測的主觀性與效率低下、自動光學檢測的高誤報與誤檢率，以及自動X 光檢測設備的昂貴與耗時，都構成了顯著挑戰。為此，本研究旨在開發一套基於邊緣運算與深度學習的PCB 電路板瑕疵即時檢測系統，以突破現有方法的瓶頸。此核心在於整合多項先進技術。首先，影像預處理用以優化影像品質並強化瑕疵特徵。使用深度學習模型作為系統核心，提供快速且精確的識別功能。為應對高解析度PCB 影像中的微小瑕疵，引入了切片輔助超推論技術。同時，為確保系統能在邊緣裝置上高效運行，模型剪枝被應用於精簡演算法的參數與計算需求，進而提升推論速度與效率。影像預處理技術有效提升模型對多樣化輸入的適應性與偵測性能。模型剪枝在維持高準確率的同時，成功降低模型耗能與推論時間，證實在資源有限的邊緣裝置上具備高度可行性。SAHI 技術顯著提升了模型對微小瑕疵的偵測能力及預測信心分數，是強化整體檢測準確度的關鍵方案。最終，成功將整套系統部署於樹梅派5，充分發揮了運算低延遲、高即時性、降低雲端依賴及成本效益等優勢。這項研究成果不僅證明提出的技術組合能有效提升邊緣裝置上AI 影像檢測的效能，為智慧製造領域的PCB 瑕疵即時檢測提供了一套高效且可靠的創新解決方案。

Printed Circuit Boards (PCBs) play a core role in modern electronic products, and their defect detection is crucial for ensuring product quality and reliability. Traditional detection methods such as manual inspection and Automated Optical Inspection (AOI) face challenges like subjectivity, low efficiency, high false positive rates, and false negative rates, while Automated X-ray Inspection (AXI) equipment is expensive and time-consuming. To address these limitations, this study aims to design and implement a real-time PCB defect detection system based on edge computing and deep learning.This research integrates image preprocessing techniques, advanced object detection algorithms, SAHI technology, and Model Pruning technology. Image preprocessing (Non-Local Means denoising, CLAHE, sharpening, and morphological closing) is used to enhance image quality and highlight defect features. A deep learning model serves as the core detector, providing fast and accurate identification capabilities. SAHI technology performs sliced inference on high-resolution PCB images, improving the ability to detect tiny defects. Model Pruning technology is employed.to reduce algorithm parameters and computational resource requirements, thereby increasing inference speed and efficiency on edge devices.Experimental results show that image preprocessing effectively enhances the model's adaptability to different types of input and improves detection performance. Model pruning significantly reduces model parameters and inference time while maintaining high accuracy, validating its feasibility on resource-constrained edge devices. SAHI technology successfully improves the model's ability to detect tiny defects and increases confidence scores, serving as an effective auxiliary method for enhancing overall detection accuracy.This study successfully implemented the system on a Raspberry Pi 5 edge device, fully leveraging the advantages of edge computing, such as low latency, high real-time performance, reduced reliance on the cloud, and cost-effectiveness. The research findings demonstrate that the proposed combination of techniques can effectively improve the performance of AI image detection on edge devices, providing an efficient and reliable solution for real-time PCB defect detection in the smart manufacturing domain.

[1] B. Du, F. Wan, G. Lei, L. Xu, C. Xu, and Y. Xiong, “Yolo-mbbi: Pcb surface defect detection method based on enhanced yolov5,” Electronics, vol. 12, no. 13, p. 2821, 2023.
[2] 郭振雄, “一种tft-lcd 电极线路缺陷检测的aoi 设备控制系统的研发与设计,” Master’s thesis, 厦门大学, 2017.
[3] 苏佳, 贾欣雨, and 侯卫民, “基于yolo-j 的pcb 缺陷检测算法,” 计算机集成制造系统, vol. 30, no. 11, p. 3984, 2024.
[4] M. Deepa, M. Kaniskaa, K. Hariprasath, S. Eashwar, and M. Subbu, “Pcb component inspection system using raspberry pi,” in 2023 IEEE Technology & Engineering Management Conference-Asia Pacific (TEMSCON-ASPAC). IEEE, 2023, pp. 1–6.
[5] L. Zhu, S. Wang, M. Chen, Y. Zhu, K. Xing, and A. Shen, “Subtle defect detection network: More accurately detect subtle defects on the printed circuit board surface,” Engineering Applications of Artificial Intelligence, vol. 152, p. 110765, 2025.
[6] J. Wang, X. Xie, G. Liu, and L. Wu, “A lightweight pcb defect detection algorithm based on improved yolov8-pcb,” Symmetry, vol. 17, no. 2, p. 309, 2025.
[7] Y. Zhou, M. Yuan, J. Zhang, G. Ding, and S. Qin, “Review of vision-based defect detection research and its perspectives for printed circuit board,” Journal of Manufacturing Systems, vol. 70, pp. 557–578, 2023.
[8] S. Zhuo, J. Shi, X. Zhou, and J. Kan, “Yolov7-tid: A lightweight network for pcb intelligent detection,” IEEE Access, 2024.
[9] K. Zhu, F. Hu, Y. Ding, W. Zhou, and R. Wang, “A comprehensive review of network pruning based on pruning granularity and pruning time perspectives,” Neurocomputing, p. 129382, 2025.
[10] M. Abd Al Rahman and A. Mousavi, “A review and analysis of automatic optical inspection and quality monitoring methods in electronics industry,” Ieee Access, vol. 8,pp. 183 192–183 271, 2020.
[11] R. Pramudita and F. I. Hariadi, “Development of techniques to determine object shifts for pcb board assembly automatic optical inspection (aoi),” in 2018 International symposium on electronics and smart devices (ISESD). IEEE, 2018, pp. 1–4.
[12] J. Shim, H. S. Cho, and S. Kim, “A new probing system for the in-circuit test of a pcb,”in Proceedings of IEEE International Conference on Robotics and Automation, vol. 1.IEEE, 1996, pp. 590–595.
[13] R. Rotar, S. L. Jurj, N.-C. Rohatinovici, R. Brîncovan, F. Opriţoiu, and M. Vlăduţiu,“Improved flying probe-inspired in-circuit tester for practical laboratory activities,” in 2023 IEEE 29th International Symposium for Design and Technology in Electronic Packaging (SIITME). IEEE, 2023, pp. 34–40.
[14] V. A. Adibhatla, J.-S. Shieh, M. F. Abbod, H.-C. Chih, C.-C. Hsu, and J. Cheng, “Detecting defects in pcb using deep learning via convolution neural networks,” in 2018 13th International Microsystems, Packaging, Assembly and Circuits Technology Conference (IMPACT). IEEE, 2018, pp. 202–205.
[15] V. A. Adibhatla, H.-C. Chih, C.-C. Hsu, J. Cheng, M. F. Abbod, and J.-S. Shieh, “Defect detection in printed circuit boards using you-only-look-once convolutional neural networks,” Electronics, vol. 9, no. 9, p. 1547, 2020.
[16] Z. Yan, H. Zhang, R. Piramuthu, V. Jagadeesh, D. DeCoste, W. Di, and Y. Yu, “Hd-cnn: hierarchical deep convolutional neural networks for large scale visual recognition,” in Proceedings of the IEEE international conference on computer vision, 2015, pp. 2740–2748.
[17] S. Mohapatra, A. Kabra, D. Gowda, S. S. Gaonkar, and S. Sadhukha, “Pcb defect detection using cnn-based deep learning,” in International Conference on Soft Computing for Security Applications. Springer, 2023, pp. 363–376.
[18] J. Zheng, X. Sun, H. Zhou, C. Tian, and H. Qiang, “Printed circuit boards defect detection method based on improved fully convolutional networks,” IEEE Access, vol. 10, pp. 109 908–109 918, 2022.
[19] T. S. Chi, M. N. Ab Wahab, A. S. A. Mohamed, M. H. M. Noor, K. B. Kang, L. L. Chuan,and L. W. J. Brigitte, “Enhancing efficientnet-yolov4 for integrated circuit detection on printed circuit board (pcb),” IEEE Access, vol. 12, pp. 25 066–25 078, 2024.
[20] S. A. Althubiti, F. Alenezi, S. Shitharth, S. K, and C. V. S. Reddy, “Circuit manufacturing defect detection using vgg16 convolutional neural networks,” Wireless Communications and Mobile Computing, vol. 2022, no. 1, p. 1070405, 2022.
[21] A. Bhattacharya and S. G. Cloutier, “End-to-end deep learning framework for printed circuit board manufacturing defect classification,” Scientific reports, vol. 12, no. 1, p.12559, 2022.
[22] J. Redmon and A. Farhadi, “Yolov3: An incremental improvement,” arXiv preprintarXiv:1804.02767, 2018.
[23] S. Carata, R. Mihaescu, E. Barnoviciu, M. Chindea, M. Ghenescu, and V. Ghenescu,“Complete visualisation, network modeling and training, web based tool, for the yolo deep neural network model in the darknet framework,” in 2019 IEEE 15th International Conference on Intelligent Computer Communication and Processing (ICCP). IEEE,2019, pp. 517–523.
[24] J. Zhong, H. Qian, H. Wang, W. Wang, and Y. Zhou, “Improved real-time object detection method based on yolov8: a refined approach,” Journal of Real-Time Image Processing,vol. 22, no. 1, pp. 1–13, 2025.
[25] R. Varghese and M. Sambath, “Yolov8: A novel object detection algorithm with enhanced performance and robustness,” in 2024 International Conference on Advances in Data Engineering and Intelligent Computing Systems (ADICS). IEEE, 2024, pp.1–6.
[26] H. Khorsand, S. Arezoomandan, and D. K. Han, “Enhanced long-range uav detection:Leveraging slicing aided hyper inference with yolov8,” in 2025 IEEE International Conference on Consumer Electronics (ICCE). IEEE, 2025, pp. 1–6.
[27] F. C. Akyon, S. O. Altinuc, and A. Temizel, “Slicing aided hyper inference and finetuning for small object detection,” in 2022 IEEE international conference on image processing (ICIP). IEEE, 2022, pp. 966–970.
[28] G. Li, X. Liao, X. Su, and B. Chao, “Pcb defect detection based on improved yolox al-gorithm,” in 2023 5th International Conference on Electrical Engineering and Control Technologies (CEECT). IEEE, 2023, pp. 380–384.
[29] O. Best, A. Khan, M. Gianni, S. Sharma, and K. Collins, “Wind turbine surface damage detection using yolov7 with slicing aided hyper inference (sahi),” in Proceedings of SAI Intelligent Systems Conference. Springer, 2023, pp. 564–576.
[30] W. Abebe, S. Jafari, S. Yu, A. Dutta, J. Strube, N. R. Tallent, L. Guo, P. Munoz, and A. Jannesari, “Supersam: Crafting a sam supernetwork via structured pruning and unstructured parameter prioritization,” arXiv preprint arXiv:2501.08504, 2025.
[31] S. Anwar, K. Hwang, and W. Sung, “Structured pruning of deep convolutional neural networks,” ACM Journal on Emerging Technologies in Computing Systems (JETC), vol. 13, no. 3, pp. 1–18, 2017.
[32] H. Ko, J.-K. Kang, and Y. Kim, “An efficient and fast filter pruning method for object detection in embedded systems,” in 2024 IEEE 6th International Conference on AI Circuits and Systems (AICAS). IEEE, 2024, pp. 204–207.
[33] Y. Cai, W. Hua, H. Chen, G. E. Suh, C. De Sa, and Z. Zhang, “Structured pruning is all you need for pruning cnns at initialization,” arXiv preprint arXiv:2203.02549, 2022.
[34] J. Zhang, P. Wang, Z. Zhao, and F. Su, “Pruned-yolo: Learning efficient object detector using model pruning,” in International conference on artificial neural networks.Springer, 2021, pp. 34–45.
[35] Z. Yuan, X. Tang, H. Ning, and Z. Yang, “Lw-yolo: Lightweight deep learning model for fast and precise defect detection in printed circuit boards,” Symmetry, vol. 16, no. 4,p. 418, 2024.
[36] L. Cai and J. Li, “Pcb defect detection system based on image processing,” in Journal of Physics: Conference Series, vol. 2383, no. 1. IOP Publishing, 2022, p. 012077.
[37] F. Zhang, N. Qiao, and J. Li, “A pcb photoelectric image edge information detection method,” Optik, vol. 144, pp. 642–646, 2017.
[38] X. Zhang, “Two-step non-local means method for image denoising,” Multidimensional Systems and Signal Processing, vol. 33, no. 2, pp. 341–366, 2022.
[39] K. Leng, “An improved non-local means algorithm for image denoising,” in 2017 IEEE 2nd International Conference on Signal and Image Processing (ICSIP). IEEE, 2017,pp. 149–153.
[40] J. Ma, X. Fan, S. X. Yang, X. Zhang, and X. Zhu, “Contrast limited adaptive histogram equalization-based fusion in yiq and hsi color spaces for underwater image enhancement,”International Journal of Pattern Recognition and Artificial Intelligence, vol. 32,no. 07, p. 1854018, 2018.
[41] V. Stimper, S. Bauer, R. Ernstorfer, B. Schölkopf, and R. P. Xian, “Multidimensional contrast limited adaptive histogram equalization,” IEEE Access, vol. 7, pp. 165 437–165 447, 2019.
[42] J. L. Clark, C. P. Wadhwani, K. Abramovitch, D. D. Rice, and M. T. Kattadiyil, “Effectof image sharpening on radiographic image quality,” The Journal of prosthetic dentistry,vol. 120, no. 6, pp. 927–933, 2018.
[43] M. Raj, A. Kumar, S. Bansal, N. Singh, M. Jain, and S. Banoth, “Accurate brain tumor detection using laplacian based gan model,” in 2025 3rd International Conference on Disruptive Technologies (ICDT). IEEE, 2025, pp. 1033–1038.
[44] K. A. M. Said, A. B. Jambek, and N. Sulaiman, “A study of image processing using morphological opening and closing processes,” International Journal of Control Theory and Applications, vol. 9, no. 31, pp. 15–21, 2016.
[45] S. Hou, C. Jia, K. Li, L. Fan, J. Guo et al., “Edge detection of covid-19 ct image based on gf_ssr, improvedmultiscale morphology, and adaptive threshold.” CMES-Computer Modeling in Engineering & Sciences, vol. 132, no. 1, 2022.
[46] A. Roy, M. M. Al Hasan, S. Ghosh, N. Varshney, J. Julia, R. Forghani, and N. Asadizanjani,“Applications and challenges of ai in pcb x-ray inspection: A comprehensive study,” ACM Journal on Emerging Technologies in Computing Systems, vol. 21, no. 1,pp. 1–28, 2025.
[47] M. Chitra, H. Balasubramanian, V. L. Srinivaas, and K. Jayanth, “An adaptation and evaluation for object detection of small objects using yolo8,” in 2024 International Conference on Intelligent Computing and Emerging Communication Technologies (ICEC).IEEE, 2024, pp. 1–9.
[48] J. Ye, Y. Wu, and W. Rong, “Based on the optimization and performance evaluation of yolov8 object detection model with multi-backbone network fusion,” in 2024 IEEE International Conference on Mechatronics and Automation (ICMA). IEEE, 2024, pp.269–274.
[49] H. Wang, S. Shen, and M. Li, “Pcb defect detection algorithm based on improved yolov8,” in 2024 5th International Conference on Electronic Communication and Artificial Intelligence (ICECAI). IEEE, 2024, pp. 323–327.
[50] R. Ye, Q. Gao, Y. Qian, J. Sun, and T. Li, “Improved yolov8 and sahi model for the collaborative detection of small targets at the micro scale: A case study of pest detection in tea,” Agronomy, vol. 14, no. 5, p. 1034, 2024.86