Convolutional neural networks, with their powerful feature extraction capabilities, provide a solid foundation for accurately identifying and tracking targets. Although convolutional neural networks have made great progress in the field of image recognition and classification, in order to further improve the accuracy and robustness of tracking, this paper proposes a multi-target tracking method based on convolutional neural networks and fractal analysis, and constructs an efficient multi-target tracking system. The paper first effectively captures target feature information by learning and training a large amount of data, and then combines the feature description of fractal geometry to mine hidden fractal features from complex data to better respond to target morphological changes and environmental interference. Experimental results show that this method can achieve significant improvements in target tracking accuracy and robustness, especially in dense and high-dynamic scenes, and can achieve stable and accurate tracking of multiple targets, providing new ideas and methods for the field of multi-target tracking.