簡易檢索 / 詳目顯示

研究生: 黃品魁
Huang, Pin-Kuei
論文名稱: 基於神經網路與地域分析的多攝影機人物追蹤分析
Multi-camera Person Tracking Analysis Based on Neural Network and Regional Analysis
指導教授: 李忠謀
Lee, Chung-Mou
口試委員: 李忠謀
Lee, Chung-Mou
江政杰
Chiang, Cheng-Chieh
柯佳伶
Koh, Jia-Ling
劉寧漢
Liu, Ning-Han
口試日期: 2023/07/20
學位類別: 碩士
Master
系所名稱: 資訊工程學系
Department of Computer Science and Information Engineering
論文出版年: 2023
畢業學年度: 111
語文別: 中文
論文頁數: 33
中文關鍵詞: 人物偵測人物追蹤人物重識別地域分析
英文關鍵詞: Person detection, Person tracking, Person re-identification, geospatial analysis
研究方法: 實驗設計法
DOI URL: http://doi.org/10.6345/NTNU202301272
論文種類: 學術論文
相關次數: 點閱:107下載:0
分享至:
查詢本校圖書館目錄 查詢臺灣博碩士論文知識加值系統 勘誤回報
  • 行人重識別技術旨在實現在不同攝影機和多個幀中跨攝影鏡頭追蹤和匹配人物的能力。該技術的應用非常廣泛,包括法務機關的嫌疑人追蹤、體育賽事的分析以及博物館人流分析等。然而,現有的重識別方法存在特徵提取和比對的限制,需要更多的計算資源且準確性有限。因此,研究人員致力於提出更準確、高效的方法,例如基於神經網路和地域分析的多攝影機人物追蹤分析方法,以改善人物重識別的準確性和實用性。
    在人物辨識與追蹤方面,本研究使用了Yolov7來判斷人物位置,並使用StrongSORT技術來進行人物匹配追蹤。而為了提高重識別的速度與準確性,本研究設計了消失出現區域和視角消失出現時間兩種地域因素來限縮匹配範圍,並設計一個檢查機制,減少匹配錯誤的可能性。預期透過地域分析的幫助下,能給予現有的追蹤與重識別技術提供更精準的結果。
    在實驗中,測試了地域分析下的OSNet和ResNet101重識別模型,相較直接對於全部的ID進行重識別的方法提高了約25%的準確度。特別是在地域分析的IDFR指標提高了29%的準確度,同時減少了運算壓力和時間,顯示地域分析演算法在實際應用中具有提升人物追蹤的效果,可帶來更準確、高效的人物追蹤和重識別解決方案。

    Person re-identification technology aims to enable the ability to track and match individuals across different cameras and multiple frames. This technology has wide-ranging applications, including suspect tracking for law enforcement agencies, analysis of sports events, and crowd analysis in museums, among others. However, existing re-ID methods have limitations in feature extraction and matching, requiring more computational resources and having limited accuracy. Therefore, researchers are devoted to proposing more accurate and efficient approaches, such as neural network-based methods and geospatial analysis, to improve the accuracy and practicality of person re-identification.
    In this research, YOLOv7 was employed for person detection to determine the positions of individuals, and the StrongSORT technique was utilized for person matching and tracking. In order to enhance the speed and accuracy of reidentification, two geographical factors were introduced: the concept of vanishing and reappearing regions and the perspective's vanishing and reappearing time. These factors were utilized to narrow down the matching scope. Additionally, a verification mechanism was designed to reduce the possibility of matching errors. It is expected that with the aid of geographical analysis, more precise results can be achieved for the existing tracking and reidentification techniques.
    In the experiment, the OSNet (Omni-Scale Network) and ResNet101 recognition models were tested under geographical analysis. Compared to the direct recognition method on all IDs, they achieved an improvement of approximately 25% in accuracy. Particularly, the IDFR indicator in geographical analysis showed a 29% increase in accuracy, while reducing computational load and time. These results demonstrate that geographical analysis algorithms have the potential to enhance person recognition in practical applications, providing more accurate and efficient solutions for person tracking and recognition.

    第一章 緒論 1 1.1 研究動機 1 1.2 研究目的 2 1.3 論文架構 2 第二章 文獻探討 3 2.1 人物偵測方法 3 2.1.1 基於R-CNN的方法 3 2.1.2 基於回歸和分類模型的方法 4 2.2 多人物追蹤方法 5 2.2.1 CenterTrack 5 2.2.2 基於骨架的追蹤 6 2.2.3 基於辨識的跟蹤 6 2.3 人物重識別方法 7 2.3.1 SSG 8 2.3.2 PCB with RPP 8 2.3.3 OSNet 8 第三章 研究方法 10 3.1 研究架構 10 3.2 人物辨識與追蹤 11 3.3 地域分析 12 3.3.1 消失出現區域 13 3.3.2 視角消失出現時間 14 3.3.3 檢查機制 15 3.4 跨攝影機人物重識別 16 第四章 實驗結果與討論 17 4.1 實驗環境與影像蒐集 17 4.2 成果量化指標 20 4.3 實驗量化結果 21 4.4 實驗結果與討論 22 4.4.1 對照組:無地域分析下的重識別 22 4.4.2 實驗組:地域分析下的重識別 24 第五章 結論與未來展望 28 5.1 結論 28 5.2 未來展望 29 參考文獻 30

    A. Bewley, Z. Ge, L. Ott, F. Ramos, and B. Upcroft, “Simple online and realtime tracking,”CoRR, vol. abs/1602.00763, 2016. [Online]. Available: http://arxiv.org/abs/1602.00763
    C. -C. Huang and M. H. Nguyen, "Robust 3D Skeleton Tracking based on OpenPose and a Probabilistic Tracking Framework," 2019 IEEE International Conference on Systems, Man and Cybernetics (SMC), Bari, Italy, 2019, pp. 4107-4112, doi: 10.1109/SMC.2019.8913977.
    C. Liu, W. Wang, H. Liu and J. Wang, "Application of Hawk-Eye Technology to Sports Events," 2022 2nd International Conference on Information Technology and Contemporary Sports (TCS), Guangzhou, China, 2022, pp. 1-5, doi: 10.1109/TCS56119.2022.9918811.
    C.-Y. Wang, A. Bochkovskiy and H.-Y. M. Liao, YOLOv7: Trainable bag-of-freebies sets new state-of-the-art for realtime object detectors, arXiv preprint arXiv:2207.02696, 2022.
    C. -J. Yu, C. -A. Chen and S. -L. Chen, "Bio-medical Image Analysis for Diagnosis and Healthcare Detection System of Skin Cancer," 2021 IEEE International Conference on Consumer Electronics-Taiwan (ICCE-TW), Penghu, Taiwan, 2021, pp. 1-2, doi: 10.1109/ICCE-TW52618.2021.9603067.
    J. Ashida, R. Miyamoto, H. Tsutsui, T. Onoye and Y. Nakamura, "Probabilistic Pedestrian Tracking Based on a Skeleton Model," 2006 International Conference on Image Processing, Atlanta, GA, USA, 2006, pp. 2825-2828, doi: 10.1109/ICIP.2006.312996.
    J. Redmon et al., “You only look once: Unified, real-time object detection,” in Proc. CVPR, pp. 779–788, 2016.
    K. He et al., “Mask R-CNN,” in Proc. ICCV, pp. 2980–2988, 2017.
    K. Zhou, Y. Yang, A. Cavallaro and T. Xiang, "Omni-Scale Feature Learning for Person Re-Identification," 2019 IEEE/CVF International Conference on Computer Vision (ICCV), Seoul, Korea (South), 2019, pp. 3701-3711, doi: 10.1109/ICCV.2019.00380.
    Li, B., Chen, H., Chen, Y., Dai, Y., He, M., "Skeleton boxes: Solving skeleton based action detection with a single deep convolutional neural network", IEEE International Conference on Multimedia and Expo Workshops, pp. 613–616, 2017.
    M. Cruz, J. J. Keh, R. Deticio, C. V. Tan, J. A. Jose and E. Dadios, "A People Counting System for Use in CCTV Cameras in Retail," 2020 IEEE 12th International Conference on Humanoid, Nanotechnology, Information Technology, Communication and Control, Environment, and Management (HNICEM), Manila, Philippines, 2020, pp. 1-6, doi: 10.1109/HNICEM51456.2020.9400048.
    Napolean, Yeshwanth & Wibowo, Priadi & Gemert, Jan. (2019). Running Event Visualization using Videos from Multiple Cameras. 82-90. 10.1145/3347318.3355528.
    N. Wojke, A. Bewley, and D. Paulus, “Simple online and realtime tracking with
    a deep association metric,” CoRR, vol. abs/1703.07402, 2017. [Online]. Available:
    http://arxiv.org/abs/1703.
    N. Gheissari, T. B. Sebastian, and R. Hartley, “Person reidentification using spatiotemporal appearance,” in 2006 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR’06), vol. 2. IEEE, 2006, pp. 1528–1535.
    Pereira R, Carvalho G, Garrote L, Nunes UJ. Sort and Deep-SORT Based Multi-Object Tracking for Mobile Robotics: Evaluation with New Data Association Metrics. Applied Sciences. 2022; 12(3):1319. https://doi.org/10.3390/app12031319
    R. Girshick, “Fast R-CNN,” in Proc. ICCV, pp. 1440-1448, 2015.
    R. Girshick, J. Donahue, T. Darrell and J. Malik, "Rich Feature Hierarchies for Accurate Object Detection and Semantic Segmentation," In 2014 IEEE Conference on Computer Vision and Pattern Recognition, pp. 580-587, 2014
    Ren, Shaoqing and He, Kaiming and Girshick, Ross and Sun, Jian., "Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks, " in Proc. NIPS pp. 91–99, 2015.
    R. Siv, M. Mancas, S. Sreng, S. Chhun and B. Gosselin, "People Tracking and Re-Identifying in Distributed Contexts: PoseTReID Framework and Dataset," 2020 12th International Conference on Information Technology and Electrical Engineering (ICITEE), Yogyakarta, Indonesia, 2020, pp. 323-328, doi: 10.1109/ICITEE49829.2020.9271712.
    Sun, Yifan & Zheng, Liang & Yang, Yi & Tian, Qi & Wang, Shengjin. (2018). Beyond Part Models: Person Retrieval with Refined Part Pooling (and A Strong Convolutional Baseline): 15th European Conference, Munich, Germany, September 8-14, 2018, Proceedings, Part IV. 10.1007/978-3-030-01225-0_30.\nW. Liu et al., “SSD: Single shot multibox detector,” in Proc. ECCV, pp. 21–37, 2016.
    X. M. Bojarski, D. D. Testa, D. Dworakowski, B. Firner, B. Flepp, P. Goyal, L. D.
    Jackel, M. Monfort, U. Muller, J. Zhang, X. Zhang, J. Zhao, and K. Zieba, “End to
    end learning for self-driving cars,” CoRR, vol. abs/1604.07316, 2016.
    Y. Du et al., "StrongSORT: Make DeepSORT Great Again," in IEEE Transactions on Multimedia, doi: 10.1109/TMM.2023.3240881.
    Y. Zhang, C. Wang, X. Wang, W. Zeng, and W. Liu, “A simple baseline for multi-object tracking,” CoRR, vol. abs/2004.01888, 2020. [Online]. Available: https://arxiv.org/abs/2004.01888
    Y. Fu, et al., "Self-Similarity Grouping: A Simple Unsupervised Cross Domain Adaptation Approach for Person Re-Identification," in 2019 IEEE/CVF International Conference on Computer Vision (ICCV), Seoul, Korea (South), 2019 pp. 6111-6120.
    Z. Cao, T. Simon, S.-E. Wei and Y. Sheikh, "Realtime multi-person 2d pose estimation using part affinity fields", IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 1302-1310, 2017.
    Z. Ge, S. Liu, F. Wang, Z. Li, and J. Sun, “YOLOX: exceeding YOLO series in 2021,” CoRR, vol. abs/2107.08430, 2021.
    Z. Lu, V. Rathod, R. Votel, and J. Huang, “Retinatrack: Online single stage joint detection and tracking,” CoRR, vol. abs/2003.13870, 2020. [Online]. Available: https://arxiv.org/abs/2003.13870
    Z. -Q. Zhao, P. Zheng, S. -T. Xu and X. Wu, "Object Detection With Deep Learning: A Review," in IEEE Transactions on Neural Networks and Learning Systems, vol. 30, no. 11, pp. 3212-3232, Nov. 2019.
    Zhou, Xingyi & Koltun, Vladlen & Krähenbühl, Philipp. (2020). Tracking Objects as Points. 10.1007/978-3-030-58548-8_28.

    無法下載圖示 電子全文延後公開
    2027/08/04
    QR CODE