隨著半導體製程持續朝奈米尺度發展，極紫外光微影技術已成為先進晶片製造的重要核心。極紫外光光罩盒作為光罩於生產、儲存及運輸過程中的保護容器，其表面品質與結構完整性將直接影響光罩潔淨度及晶圓製程良率。然而，光罩盒在製造與長期使用過程中，容易因機械震動、靜電附著或人為操作而產生刮痕、汙點及凹陷等微小瑕疵。傳統人工檢測方式不僅耗時，亦容易受到檢測人員主觀判斷及環境光線變化影響，進而造成漏檢與誤判。
本研究提出一套應用於極紫外光光罩盒之智慧化表面瑕疵檢測系統，針對高解析度工業影像中的微小瑕疵進行自動辨識。由於原始影像解析度極高，且瑕疵尺寸相對微小，若直接將完整影像輸入深度學習模型進行偵測，容易造成記憶體使用量過高，並可能因影像縮放而導致微小物件特徵流失。因此，本研究採用分塊式影像切割策略，將大型影像切分為多個較小的影像區塊進行偵測，再透過座標還原及非極大值抑制機制，整合各影像區塊的偵測結果，以完成原始高解析度影像之瑕疵定位。
在模型設計方面，本研究導入適用於微小物件辨識之深度學習物件偵測模型，以提升微小瑕疵的辨識能力。實驗結果顯示，本研究所提出之方法能有效提升高解析度工業影像中的微小瑕疵檢測效能，並兼顧檢測準確性與實際部署需求。

As semiconductor manufacturing processes continue to advance toward the nanometer scale, Extreme Ultraviolet Lithography (EUV) has become a critical technology for advanced integrated circuit fabrication. The EUV mask pod serves as a protective container for photomasks during production, storage, and transportation. Its surface quality and structural integrity directly affect mask cleanliness and wafer fabrication yield. However, during manufacturing and long-term usage, mask pods are susceptible to various surface defects, such as scratches, stains, and pits, caused by mechanical vibration, electrostatic contamination, or human handling. Traditional manual inspection methods are not only time-consuming but also prone to subjective judgment and environmental lighting variations, leading to missed detections and false alarms.
This study proposes an intelligent surface defect inspection system for EUV mask pods, aiming to automatically identify tiny defects in high-resolution industrial images. Due to the extremely high resolution of the acquired images and the small size of the target defects, directly applying deep learning models often results in excessive memory consumption and the loss of small-object features. To address these challenges, a tile-based image partitioning strategy is adopted, in which large images are divided into multiple smaller patches for defect detection. The detection results are then reconstructed through coordinate restoration and integrated using a Non-Maximum Suppression (NMS) mechanism to generate the final defect predictions on the original large-scale images.
Furthermore, a deep learning-based small object detection model is employed to enhance the recognition capability of tiny defects. Experimental results demonstrate that the proposed method effectively improves the detection performance of small defects in high-resolution industrial images while maintaining both high detection accuracy and practical deployment feasibility.

[1] D. Kazazis et al., “Extreme ultraviolet lithography,” Nature Reviews Methods Primers, vol. 4, Art. no. 84, 2024, doi: 10.1038/s43586-024-00361-z.

[2] L. He et al., “Protection efficiency of a standard compliant EUV reticle handling solution,” in Proc. SPIE 7271, Alternative Lithographic Technologies, Art. no. 72710I, 2009, doi: 10.1117/12.814304.

[3] Abd Al Rahman Abu Ebayyeh and Alireza Mousavi, “A review and analysis of automatic optical inspection and quality monitoring methods in electronics industry,” IEEE Access, vol. 8, pp. 183192–183271, 2020, doi: 10.1109/ACCESS.2020.3029127.

[4] Yajun Chen, Yuanyuan Ding, Fan Zhao, Erhu Zhang, Zhangnan Wu, and Linhao Shao, “Surface defect detection methods for industrial products: A review,” Applied Sciences, vol. 11, no. 16, Art. no. 7657, 2021, doi: 10.3390/app11167657.

[5] Rafael C. Gonzalez and Richard E. Woods, Digital Image Processing, 4th ed. New York, NY, USA: Pearson, 2018.

[6] Yann LeCun, Yoshua Bengio, and Geoffrey Hinton, “Deep learning,” Nature, vol. 521, no. 7553, pp. 436–444, 2015, doi: 10.1038/nature14539.

[7] Alex Krizhevsky, Ilya Sutskever, and Geoffrey Everest Hinton, “ImageNet classification with deep convolutional neural networks,” in Advances in Neural Information Processing Systems 25, 2012, pp. 1097–1105.

[8] Zhong-Qiu Zhao, Peng Zheng, Shou-Tao Xu, and Xindong Wu, “Object detection with deep learning: A review,” IEEE Transactions on Neural Networks and Learning Systems, vol. 30, no. 11, pp. 3212–3232, 2019, doi: 10.1109/TNNLS.2018.2876865.

[9] Ross Girshick, Jeff Donahue, Trevor Darrell, and Jitendra Malik, “Rich feature hierarchies for accurate object detection and semantic segmentation,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2014, pp. 580–587.
[10] Ross Girshick, “Fast R-CNN,” in Proceedings of the IEEE International Conference on Computer Vision (ICCV), 2015, pp. 1440–1448, doi: 10.1109/ICCV.2015.169.

[11] Shaoqing Ren, Kaiming He, Ross Girshick, and Jian Sun, “Faster R-CNN: Towards real-time object detection with region proposal networks,” in Advances in Neural Information Processing Systems 28, 2015, pp. 91–99.

[12] Jasper Uijlings, Koen van de Sande, Theo Gevers, and Arnold Smeulders, “Selective search for object recognition,” International Journal of Computer Vision, vol. 104, no. 2, pp. 154–171, 2013, doi: 10.1007/s11263-013-0620-5.

[13] Wei Liu, Dragomir Anguelov, Dumitru Erhan, Christian Szegedy, Scott Reed, Cheng-Yang Fu, and Alexander Berg, “SSD: Single shot MultiBox detector,” in Computer Vision – ECCV 2016, Lecture Notes in Computer Science, vol. 9905. Cham, Switzerland: Springer, 2016, pp. 21–37, doi: 10.1007/978-3-319-46448-0_2.

[14] Joseph Redmon, Santosh Divvala, Ross Girshick, and Ali Farhadi, “You only look once: Unified, real-time object detection,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016, pp. 779–788, doi: 10.1109/CVPR.2016.91.

[15] Joseph Redmon and Ali Farhadi, “YOLO9000: Better, faster, stronger,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2017, pp. 7263–7271.

[16] Joseph Redmon and Ali Farhadi, “YOLOv3: An incremental improvement,” arXiv preprint arXiv:1804.02767, 2018.

[17] Alexey Bochkovskiy, Chien-Yao Wang, and Hong-Yuan Mark Liao, “YOLOv4: Optimal speed and accuracy of object detection,” arXiv preprint arXiv:2004.10934, 2020.

[18] Zheng Ge, Songtao Liu, Feng Wang, Zeming Li, and Jian Sun, “YOLOX: Exceeding YOLO series in 2021,” arXiv preprint arXiv:2107.08430, 2021.

[19] Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan Gomez, Łukasz Kaiser, and Illia Polosukhin, “Attention is all you need,” in Advances in Neural Information Processing Systems 30, 2017, pp. 5998–6008.

[20] Nicolas Carion, Francisco Massa, Gabriel Synnaeve, Nicolas Usunier, Alexander Kirillov, and Sergey Zagoruyko, “End-to-end object detection with transformers,” in Computer Vision – ECCV 2020, Lecture Notes in Computer Science, vol. 12346. Cham, Switzerland: Springer, 2020, pp. 213–229, doi: 10.1007/978-3-030-58452-8_13.

[21] Xizhou Zhu, Weijie Su, Lewei Lu, Bin Li, Xiaogang Wang, and Jifeng Dai, “Deformable DETR: Deformable transformers for end-to-end object detection,” in Proceedings of the 9th International Conference on Learning Representations (ICLR), 2021.

[22] Feng Li, Hao Zhang, Shilong Liu, Jian Guo, Lionel Ming-Shuan Ni, and Lei Zhang, “DN-DETR: Accelerate DETR training by introducing query denoising,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2022, pp. 13619–13627.

[23] Hao Zhang, Feng Li, Shilong Liu, Lei Zhang, Hang Su, Jun Zhu, Lionel Ming-Shuan Ni, and Heung-Yeung Shum, “DINO: DETR with improved denoising anchor boxes for end-to-end object detection,” in Proceedings of the International Conference on Learning Representations (ICLR), 2023.

[24] Yian Zhao, Wenyu Lv, Shangliang Xu, Jinman Wei, Guanzhong Wang, Qingqing Dang, Yi Liu, and Jie Chen, “DETRs beat YOLOs on real-time object detection,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2024, pp. 16965–16974.

[25] Tsung-Yi Lin, Piotr Dollár, Ross Girshick, Kaiming He, Bharath Hariharan, and Serge Belongie, “Feature pyramid networks for object detection,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2017, pp. 2117–2125, doi: 10.1109/CVPR.2017.106.

[26] Zhenghua Zhang, Ling Jiang, and Qingqing Hong, “A survey of small object detection based on deep learning,” in Proceedings of the 8th IIAE International Conference on Intelligent Systems and Image Processing, 2021, pp. 45–52, doi: 10.12792/icisip2021.009.
[27] Shu Liu, Lu Qi, Haifang Qin, Jianping Shi, and Jiaya Jia, “Path aggregation network for instance segmentation,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2018, pp. 8759–8768.

[28] Özge Ünel, Burak Oğuz Özkalaycı, and Cevahir Çığla, “The power of tiling for small object detection,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, 2019, pp. 582–591, doi: 10.1109/CVPRW.2019.00084.

[29] Juan Terven, Diana-Margarita Córdova-Esparza, and Julio-Alejandro Romero-González, “A comprehensive review of YOLO architectures in computer vision: From YOLOv1 to YOLOv8 and YOLO-NAS,” Machine Learning and Knowledge Extraction, vol. 5, no. 4, pp. 1680–1716, 2023, doi: 10.3390/make5040083.

[30] Kentaro Wada, Labelme: Image Polygonal Annotation with Python, computer software, Zenodo, 2021, doi: 10.5281/zenodo.5711226.

[31] Rahima Khanam and Muhammad Hussain, “YOLOv11: An overview of the key architectural enhancements,” arXiv preprint arXiv:2410.17725, 2024.

[32] Muhammad Yaseen, “What is YOLOv8: An in-depth exploration of the internal features of the next-generation object detector,” arXiv preprint arXiv:2408.15857, 2024.

[33] Glenn Jocher, Jing Qiu, Mengyu Liu, Shuai Lyu, Fatih Cagatay Akyon, and Muhammet Esat Kalfaoglu, “Ultralytics YOLO26: Unified real-time end-to-end vision models,” arXiv preprint arXiv:2606.03748, 2026, doi: 10.48550/arXiv.2606.03748.