Deep Learning-Based Real-Time Suspicious Activity Detection System

Deep Learning-Based Suspicious Activity and Scream De- tection System from Live Camera Audio and Video Datasets The system presented in this paper proposes a surveillance monitoring solution that utilizes deep learning algorithms for analyzing audio-video datasets captured by live cameras. The system combines computer vision technology for video processing and audio processing technology for identifying abnormalities in the audio signals from the surveillance en- vironment. It uses CNN along with the YOLOv8 algorithm for video-based suspicious activity detection, while the MLP algorithm works for the audio-based scream detection task. The proposed system has proven its capability to detect abnormalities in surveillance environments in real-time by capturing video frames and audio signals and performing analysis using deep learning models. As soon as any suspicious activities such as abnormal movements or distress calls are identified, the system captures the video frames with the detected event and saves the information related to the event such as time and place of occurrence. After that, it sends an email notification to the authorized user about the detected event. As a result, the system proves to be efficient enough when it comes to detecting unusual activities and human screams and can therefore be applied in practical applications like public security systems, schools, traffic stations, smart cities, and so forth. The graphical user interface created with the help of Python Tkinter is user-friendly and allows efficient control. In regard to future development, the system can be improved by connecting several cameras in order to achieve large scale surveillance and introducing deep learning techniques for achieving better detection accuracy. Furthermore, the system can be modified in order to include mobile notifications and cloud-based storage in addition to location tracking. The future development can involve the introduction of certain types of human behavior detection as well as recognition of weapon objects and crowds.

[1]     V. Mandalapu, P. Kumar, and R. Sharma, “Crime Prediction Using Machine Learning and Deep Learning: A Systematic Review and Future Directions,” International Journal of Advanced Computer Science and Applications, vol. 14, no. 3, pp. 120–130, 2023. [2]     M. Mudgal, A. Singh, and P. Verma, “Theoretical and Empirical Anal- ysis of Crime Data Using Machine Learning Techniques,” International Journal of Computer Applications, vol. 183, no. 42, pp. 15–22, 2021. [3]     T. Saba, A. Rehman, and M. Sharif, “Suspicious Activity Recognition Using Proposed Deep L4-Branched-ActionNet With Entropy Coded Optimization,” IEEE Access, vol. 9, pp. 115234–115245, 2021. [4]     S. Loganathan, K. Prakash, and R. Kumar, “Deep Learning-Based Suspicious Activity Detection Using Video Surveillance,” International Conference on Artificial Intelligence and Signal Processing, pp. 245– 250, 2019. [5]     N. Bordoloi, R. Das, and S. Roy, “Real-Time Suspicious Activity Detection Using YOLO-Based Object Detection,” International Journal of Computer Vision Applications, vol. 12, no. 2, pp. 67–75, 2020. [6]     U. M. Kamthe and S. Patil, “Hierarchical Semantic-Based Suspicious Activity Detection Using Motion Tracking,” Proceedings of IEEE Inter- national Conference on Image Processing, pp. 98–103, 2018. [7]     J. Redmon, S. Divvala, R. Girshick, and A. Farhadi, “You Only Look Once: Unified, Real-Time Object Detection,” in Proc. IEEE Conf. on Computer Vision and Pattern Recognition (CVPR), 2016, pp. 779–788. [8]     A. Bochkovskiy, C. Wang, and H. Liao, “YOLOv4: Optimal Speed and Accuracy of Object Detection,” arXiv preprint arXiv:2004.10934, 2020. [9]     K. He, X. Zhang, S. Ren, and J. Sun, “Deep Residual Learning for Image Recognition,” in Proc. IEEE Conf. on Computer Vision and Pattern Recognition (CVPR), 2016, pp. 770–778. [10]    Y. LeCun, L. Bottou, Y. Bengio, and P. Haffner, “Gradient-Based Learning Applied to Document Recognition,” Proceedings of the IEEE, vol. 86, no. 11, pp. 2278–2324, 1998.