Multi-target Tracking Algorithm for Film and Video under Multi-Layer Convolution Feature

Accurate multi-target tracking in film and video has become increasingly important for intelligent visual perception systems and electromagnetic-enabled sensing applications, where robust signal representation and feature propagation are essential for reliable scene understanding. This study investigates the accuracy and precision of multi-target character tracking in film and video by employing a multi-layer convolution feature framework integrated with convolutional neural networks (CNNs). In the proposed neural network architecture, video data are processed through multiple convolution and pooling layers followed by fully connected layers, while the Rectified Linear Unit (ReLU) is adopted as the activation function to enhance network performance. Experimental results demonstrate that the proposed method achieves a multiple object tracking accuracy (MOTA) of 56.8%, a multiple object tracking precision (MOTP) of 79.8%, and an Identification F-score (IDF1) of 60.1%. On the MOT16 benchmark, the proposed approach improves MOTA, MOTP, and IDF1 by 1.8%, 4.0%, and 2.7%, respectively, compared with the Dynamic Memory based Attention Network (DMAN), indicating superior tracking consistency and feature discrimination across different video sequences. The results confirm that the CNN-based multi-target tracking framework significantly enhances MOTA and IDF1 while providing a reliable computational foundation for intelligent visual sensing and signal processing systems in advanced communication and electromagnetic perception environments.