本論文提出一基於室內攝影機監控環境之人員追蹤系統，以進行人員軌跡及
活動監控。本系統可以分為三個部分：影像深度學習、人員定位和人員活動辨識。
影像深度學習部分係透過 YOLO（You Only Look Once）v3 進行人員影像樣本
之訓練學習，並進行偵測應用。其次，影像中所偵測到之人員邊界框（Bounding
Box），會藉由攝影機之高度與姿態計算出人員之平面位置和坐標。此外，藉由人
員邊界框之移動量與長寬比例回歸分析進行分析，以判別人員在室內之活動；如
站立、行走、坐下等。本系統於實驗室中架設兩台攝影機，並把計算出之人員位
置存放於資料庫，透過兩台攝影機之坐標整合，即可達成兩台攝影機人員追蹤資
訊之整合。最後，本研究以一整合畫面來顯示整場域之人員定位和追蹤系統。

This thesis presents a human tracking system for indoor camera surveillance environments, which is used for the monitoring of human trajectory and activities. The proposed system is composed of three parts: deep learning human detection from surveillance cameras, human localization and human activity recognition. The surveillance camera images were processed in terms of YOLO（You Only Look Once）
v3 deep learning model to detect the human. The detected human areas in the image were identified as bounding boxes. The bounding boxes were further utilized with the camera spatial information, including height and pan and tilt angles, to obtain the human position relative to the camera floor coordinate system. Meanwhile, the variation of a specific bounding box of two adjacent image frames and the regression analysis of the height/ width ratio of a bounding box in an image were used to recognize the human activities of standing, walking and sitting. In this work, two surveillance cameras were deployed in our laboratory. The detected human locations of each camera were all recorded in a database. By registering the coordinates of two cameras, a global human tracking system cross different cameras can be achieved. Finally, an integrated graphical user interface was implemented to demonstrate the operation of the proposed human localization and tracking system.

[1] K. Yoon, Y. Song, and M. Jeon, “Multiple hypothesis tracking algorithm for multi-target multi-camera tracking with disjoint views,” IET Image Processing, vol. 12, no. 7, pp. 1175–1184, 2018.
[2] M. Byeon, S. Yun, Y. Ro, D. Jo, K. Kim, and J. Y. Choi, “Real-time scheme for 3-dimensional localizing and tracking of people in multiple camera settings,” International Conference on Control, Automation and Systems (ICCAS), pp. 239–244, 2017.
[3] Y. G. Lee, Z. Tang, and J. N. Hwang, “Online-Learning-Based Human Tracking Across Non-Overlapping Cameras,” IEEE Transactions on Circuits and Systems for Video Technology, pp. 1–1, 2017.
[4] A. T. Y. Chen, J. Fan, M. B. Abhari, and K. I. K. Wang, “A computationally efficient pipeline for camera-based indoor person tracking,” International Conference on Image and Vision Computing New Zealand (IVCNZ), pp. 1–6, 2017.
[5] K. Kim, B. Heo, M. Byeon, and J. Y. Choi, “Deep learning architecture for pedestrian 3-D localization and tracking using multiple cameras,” IEEE International Conference on Image Processing (ICIP), pp. 1147–1151, 2017.
[6] B. Bozorgtabar and R. Goecke, “MSMCT: Multi-State Multi-Camera Tracker,” IEEE Transactions on Circuits and Systems for Video Technology, pp. 1–1, 2017.
[7] Y. Waizumi, M. Omachi, and K. Tanaka, “On-Demand Color Calibration for Pedestrian Tracking in Nonoverlapping Fields of View,” IEEE Internet of Things Journal, vol. 4, no. 2, pp. 320–329, 2017.
[8] X. Luo, F. Wang, and M. Luo, “Collaborative target tracking in lopor with multi-camera,” Optik - International Journal for Light and Electron Optics, vol. 127, no. 23, pp. 11588–11598, 2016.
[9] H. Choi and M. Jeon, “Data association for non-overlapping multi-camera multi-object tracking based on similarity function,” IEEE International Conference on Consumer Electronics-Asia (ICCE-Asia), pp. 1–4, 2016.
[10] F. Fleuret, J. Berclaz, R. Lengagne, and P. Fua, “Multicamera People Tracking with a Probabilistic Occupancy Map,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 30, no. 2, pp. 267–282, 2008.
[11] S. C. Agrawal, R. K. Tripathi, and A. S. Jalal, “Human-fall detection from an indoor video surveillance,” International Conference on Computing, Communication and Networking Technologies (ICCCNT), pp. 1–5, 2017.
[12] M. Babiker, O. O. Khalifa, K. K. Htike, A. Hassan, and M. Zaharadeen, “Automated daily human activity recognition for video surveillance using neural network,” IEEE 4th International Conference on Smart Instrumentation, Measurement and Application (ICSIMA), pp. 1–5, 2017.
[13] M. I. Zul, I. Muslim, and L. Hakim, “Human Activity Recognition by Using Nearest Neighbor Algorithm from Digital Image,” International Conference on Soft Computing, Intelligent System and Information Technology (ICSIIT), pp. 58–61, 2017.
[14] M. Z. Uddin, W. Khaksar, and J. Torresen, “Human activity recognition using robust spatiotemporal features and convolutional neural network,” IEEE International Conference on Multisensor Fusion and Integration for Intelligent Systems (MFI), pp. 144–149, 2017.
[15] H. Samir, H. Abdelmunim, and G. M. Aly, “Human activity recognition using shape moments and normalized fourier descriptors,” International Conference on Computer Engineering and Systems (ICCES), pp. 359–364, 2017.
[16] N. Kase, M. Babaee, and G. Rigoll, “Multi-view human activity recognition using motion frequency,” IEEE International Conference on Image Processing (ICIP), pp. 3963–3967, 2017.
[17] M. Papakostas, T. Giannakopoulos, F. Makedon, and V. Karkaletsis, “Short-Term Recognition of Human Activities Using Convolutional Neural Networks,” International Conference on Signal-Image Technology & Internet-Based Systems (SITIS), pp. 302–307, 2016.
[18] M. Z. Uddin, J. Torresen, and T. Jabid, “Human Activity Recognition using depth body part histograms and Hidden Markov Models,” International Conference on Innovations in Science, Engineering and Technology (ICISET), pp. 1–4, 2016.
[19] V. Ramakrishna, T. Kanade, and Y. Sheikh, “Tracking Human Pose by Tracking Symmetric Parts,” IEEE Conference on Computer Vision and Pattern Recognition, pp. 3728–3735, 2013.
[20] M. Saito, K. Kitaguchi, H. Nishida and M. Hashimoto, “Human behavior recognition using regression models,” ICCAS-SICE, pp. 4647–4650, 2009.
[21] K. Fathian, J. P. Ramirez-Paredes, E. A. Doucette, J. W. Curtis and N. R. Gans, “QuEst: A Quaternion-Based Approach for Camera Motion Estimation From Minimal Feature Points”, IEEE Robotics and Automation Letters, pp. 857-864, 2018.
[22] S. Sengupta, T. Amir, M. Galun, T. Goldstein, D. W. Jacobs, A. Singer and R. Basri, “A New Rank Constraint on Multi-view Fundamental Matrices, and Its Application to Camera Location Recovery”, IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 2413-2421, 2017.
[23] G. Manzo, F. Serratosa and M. Vento, “Online human assisted and cooperative pose estimation of 2D cameras”, Expert Systems with Applications, pp. 258-268, 2016.
[24] I. Nurutdinova and A. Fitzgibbon, “Towards Pointless Structure from Motion: 3D Reconstruction and Camera Parameters from General 3D Curves”, IEEE International Conference on Computer Vision (ICCV), pp. 2363-2371, 2015.
[25] F. Jin and X. Wang, “An autonomous camera calibration system based on the theory of minimum convex hull”, Fifth International Conference on Instrumentation and Measurement, Computer, Communication and Control (IMCCC), 2015, pp. 857-860, 2015.
[26] F. Pirahansiah, S. N. H. S. Abdullah and S. Sahran, “Camera calibration for multi-modal robot vision based on image quality assessment”, 10th Asian Control Conference (ASCC), pp. 1-6, 2015.
[27] L. Song, W. Wu, J. Guo and X. Li, “Survey on camera calibration technique”, 5th International Conference on Intelligent Human-Machine Systems and Cybernetics, vol. 2, pp. 389-392, 2013.
[28] A. Fetić, D. Jurić and D. Osmanković, “The procedure of a camera calibration using camera calibration toolbox for MATLAB”, Proceedings of the 35th International Convention MIPRO, 2012, pp. 1752-1757 , 2012.
[29] A. Kapadia, D. Braganza, D. M. Dawson and M. L. McIntyre, “Adaptive camera calibration with measurable position of fixed features”, American Control Conference, pp. 3869-3874, 2008.
[30] http://www.dlinktw.com.tw/home/product?id=83
[31] https://zh.wikipedia.org/wiki/Qt
[32] https://kb.site5.com/databases/phpmyadmin/phpmyadmin-running-sql-
queries-on-a-database/
[33] https://www.cnblogs.com/makefile/p/YOLOv3.html
[34] https://towardsdatascience.com/yolo-v3-object-detection-53fb7d3bfe6b