Journal of Digital Information Management Weigang Zhang, Youzi Li 22 4 2024 https://10.6025/jdim/2024/22/4/130-136 https://www.dline.info/fpaper/jdim/v22i4/jdimv22i4_3.pdf This paper focuses on applying a deep learning- based intelligent video surveillance system, particularly emphasising using the YOLOv7 model for object detection. By reviewing the development of intelligent video surveillance technology, we recognize the importance of deep learning in computer vision. The structure and characteristics of the YOLOv7 model are detailed, including the input layer, backbone network layer, feature fusion layer, and output layer. To validate the model's performance, we conducted experiments on the VOC dataset, and the results show that the YOLOv7 model achieved an average detection accuracy of 0.89 on this dataset. The experimental results demonstrate the efficiency, accuracy, and robustness of the YOLOv7 model in intelligent video surveillance, providing essential references for optimizing and applying intelligent video surveillance systems.