臺灣博碩士論文加值系統

隨著人工智慧與電腦視覺技術的進步和蓬勃發展，目前許多產業，包括：金融業、製造業、醫療業、電商等，都已陸續導入人工智慧相關技術的應用。然而，我們發現員警、維安或保全人員卻鮮少能從這一波人工智慧熱潮中受益。在文獻調查中發現：為了協助警務人員於執行勤務過程中減少其生命所遭受的威脅或傷害，全美所有員警學院遵循一項教條「21 英呎法則(Tueller Drill)」。而所謂的 21 英呎法論述到：21 英呎範圍是員警與嫌疑犯可保持的最短距離；而在此最短距離下，員警才來得及(在約 1 秒半內)拔槍、射擊與制伏嫌疑犯。為了實現此一教條，本論文提出了一套結合人工智慧與電腦視覺技術的自動防衛槍塔控制系統。更具體地說，本論文提出一槍塔的硬體機構設計，並結合人工智慧與電腦視覺技術，實現槍塔移動、瞄準、自動追蹤與扳機擊發等功能。所提系統能夠即時偵測和追蹤目標人員(如：嫌疑犯)，並在目標與槍塔距離低於 21 英呎時予以警告或採取輔助的積極行為(如：擊發扳機)。我們期待透過此一系統的設計與實現，提供關鍵時刻的電腦輔助方法，讓警務或維安人員能夠即時應對潛在威脅，進而補強值勤人員的反應能力和提升其安全性。

With the rapid development of artificial intelligence (AI) and computer vision technologies, many industries such as finance, manufacturing, healthcare, and e-commerce have already begun implementing AI-related applications. However, law enforcement, security personnel, and private security guards have rarely benefited from this AI trend. Our literature review revealed that to help reduce threats or harm to officers' lives, all police academies in the United States follow the "21-foot rule" (Tueller Drill). This rule states that 21 feet is the minimum safe distance between an officer and a suspect, giving the officer just enough time (about 1.5 seconds) to draw their weapon, shoot, and subdue the suspect. This thesis proposes an automated defensive gun turret control system that combines AI and computer vision technologies. We present a hardware mechanism design for a gun turret, integrated with AI and computer vision to achieve functions such as turret movement, aiming, automatic tracking, and trigger activation. The proposed system can detect and track target individuals (e.g., suspects) in real-time, issuing warnings or taking active measures (such as triggering the gun) when the distance between the target and the turret falls below 21 feet. We aim to provide a computer-assisted method for critical moments, allowing law enforcement and security personnel to respond to potential threats in real-time, thereby enhancing their reaction capabilities and improving their safety.

摘要 i
ABSTRACT ii
誌謝 iii
目錄 iv
表目錄 vi
圖目錄 vii
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 相關文獻 3
2.1 人體骨幹偵測與估算(POSE ESTIMATION) 3
2.2 多目標追蹤(MULTIPLE OBJECT TRACKING) 4
第三章 一套結合人工智慧與電腦視覺技術的自動防衛槍塔控制系統 7
3.1 系統機構與控制方法設計 7
3.1.1 槍塔之機構設計與實現 8
3.1.2 槍塔之相關馬達規格、電路設計與實現 12
3.1.3 基於人體關鍵點的自動瞄準方法 17
3.2 以人體骨幹預估之標示框為引導的DEEP SORT目標追蹤法 19
3.2.1 問題描述 19
3.2.2監督式機器學習方法 19
3.2.3以骨幹關節關鍵點為引導的人體標示框估算法 20
第四章 實驗結果與分析 23
4.1 訓練資料 23
4.2 標示框估算的實驗結果 26
4.3 標示框估算結果對DEEP SORT效能的影響實驗結果 29
4.4 軟硬體整合呈現 33
4.4.1 自動化瞄準機制 33
4.4.2 基於距離偵測實現自動防衛 35
第五章 結論與未來展望 37
參考文獻 38
作者簡介 40

[1] D. Tuller, “How Close is Too Close?,” S.W.A.T. Magazine, Mar. 1983.
[2] NVIDIA AI-IOT, "trt_pose," GitHub repository, https://github.com/NVIDIA-AI-IOT/trt_pose (accessed June 7, 2024).
[3] N. Wojke and A. Bewley, “Deep Cosine Metric Learing for Person Re-identification,” IEEE Winter Conf. on Applications of Computer Vision (WACV), pp. 748-756, 2018.
[4] Z. Cao, G. H. Martinez, T. Siman, S. Wei, and Y. A. Sheikh, “OpenPose: Realtime Mutli-Person 2D Pose Estimation using Part Affinity Fields,” IEEE Trans. Pattern Analysis and Machine Intelligence, 2019.
[5] J. Theeuwes, "Top–down and bottom–up control of visual selection," Acta Psychologica, vol. 135, no. 2, pp. 77-99, 2010.
[6] Lucas, B. D., and Kanade, T., "An Iterative Image Registration Technique with an Application to Stereo Vision," in Proceedings of the 7th International Joint Conference on Artificial Intelligence (IJCAI), 1981, pp. 674-679.
[7] Reid, D. B., "An Algorithm for Tracking Multiple Targets," in IEEE Transactions on Automatic Control, vol. 24, no. 6, pp. 843-854, December 1979.
[8] Cox, I. J., and Hingorani, S. L., "An Efficient Implementation of Reid's Multiple Hypothesis Tracking Algorithm and Its Evaluation for the Purpose of Visual Tracking," in IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 18, no. 2, pp. 138-150, February 1996.
[9] S. Ren, K. He, R. Girshick, and J. Sun, "Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks," IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 39, no. 6, pp. 1137-1149, June 2017.
[10] 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, 2016, pp. 779-788.
[11] T. Pfister, J. Charles, and A. Zisserman, "Flowing ConvNets for Human Pose Estimation in Videos," 2015 IEEE International Conference on Computer Vision (ICCV), Santiago, 2015, pp. 1913-1921.
[12] Y. Jia et al., "Caffe: Convolutional Architecture for Fast Feature Embedding," Proceedings of the 22nd ACM International Conference on Multimedia, Orlando, FL, 2014, pp. 675-678.
[13] D. E. Knuth, "Mathematical analysis of algorithms," Proceedings of the International Congress of Mathematicians, Nice, France, 1970, pp. 269-274.
[14] NVIDIA, "TensorRT: High Performance Deep Learning Inference Platform," NVIDIA Developer Documentation, 2018. Available: https://developer.nvidia.com/tensorrt.
[15] A. Bewley, Z. Ge, L. Ott, F. Ramos, and B. Upcroft, "Simple Online and Realtime Tracking," 2016 IEEE International Conference on Image Processing (ICIP), Phoenix, AZ, 2016, pp. 3464-3468.
[16] R. E. Kalman, "A New Approach to Linear Filtering and Prediction Problems," Journal of Basic Engineering, vol. 82, no. 1, pp. 35-45, 1960.
[17] H. W. Kuhn, "The Hungarian Method for the Assignment Problem," Naval Research Logistics Quarterly, vol. 2, no. 1-2, pp. 83-97, March 1955.
[18] L. Zheng, L. Shen, L. Tian, S. Wang, J. Wang, and Q. Tian, "Scalable Person Re-identification: A Benchmark," 2015 IEEE International Conference on Computer Vision (ICCV), Santiago, 2015, pp. 1116-1124.
[19] X. Chen, H. Zheng, and L. Zhang, "Cascade Matching for Real-Time Object Detection," 2014 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Columbus, OH, 2014, pp. 1197-1204.
[20] P. C. Mahalanobis, "On the Generalized Distance in Statistics," Proceedings of the National Institute of Sciences of India, vol. 2, no. 1, pp. 49-55, 1936.
[21] B. Li and L. Han, "Distance Weighted Cosine Similarity Measure for Text Classification," in Lecture Notes in Computer Science, vol. 8206, Springer, Berlin, Heidelberg, 2013, pp. 611-620.
[22] I. El Naqa and M. J. Murphy, "What is machine learning?," in Machine Learning in Radiation Oncology, Springer International Publishing, 2015, pp. 3-11.
[23] S. Weisberg, Applied Linear Regression, John Wiley & Sons, 2005.
[24] M. Belgiu and L. Drăguţ, "Random forest in remote sensing: A review of applications and future directions," ISPRS Journal of Photogrammetry and Remote Sensing, vol. 114, pp. 24-31, 2016.