Research And System Implementation Of Lightweight And Efficient Person Re-identification

Person re-identification is a technology that uses computer vision technology to retrieve whether there is a specific target in the video or image sequence captured by intelligent surveillance equipment.The research on person re-identification is generally divided into five steps,which are data acquisition,pedestrian bounding box generation,data labeling,model training and target retrieval.Aiming at the problem that most of the current person re-identification models have high complexity,large amount of parameters,and are difficult to play a role in the business scenarios with high real-time requirements,this paper proposes a pedestrian re-identification method based on multi-attribute features.Based on the unified network model training,the hierarchical features of pedestrians are obtained,which can well combine the low-level visual features and high-level semantic attributes of pedestrians.Obtaining highly discriminative pedestrian descriptors.The performance of the proposed method is significantly better than most of the person re-identification algorithms in the real application environment,and the Rank@ index of the proposed method is 5%-10% higher than the most existing feature representation methods.Compared with the traditional person re-identification model,the proposed model has lower complexity and significantly reduced the number of parameters,which is easier to converge,and the time complexity optimization is as low as O(n).It can match more than 5000 special instances within 3s,which has the potential for real-time analysis in real monitoring scenarios.Aiming at the lack of interface experimental scheme in current person re-identification research,this paper designs and implements an autonomous and convenient person reidentification system,which provides a simple and easy to use interface and a variety of customizable options.The interface system is convenient for users to operate,and can intuitively view the influence of parameters on the experimental results and visual experimental results.The customization option covers most of the parameters that will be used in the process of person re-identification experiment from training to testing.Users can customize the experiment scheme according to their own needs.Through the setting of the path configuration experimental parameters and access to the person re-identification algorithm,this system helps beginners and non-experts to avoid the complicated configuration and deployment process of running person re-identification experiments,and finally provides users with an autonomous and controllable person re-identification interface use scheme.