步态识别模型的训练方法、步态识别方法、装置及设备
By introducing preprocessing, spatial feature extraction, temporal feature extraction, and feature remapping networks into the gait recognition model, and optimizing the training process using a two-dimensional convolutional module, the problem of large model parameter count is solved, and efficient gait recognition training and recognition are achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- BEIJING JINGDONG SHANGKE INFORMATION TECH CO LTD
- Filing Date
- 2022-09-19
- Publication Date
- 2026-07-17
AI Technical Summary
Existing gait recognition models face challenges in real-world scenarios, such as occlusion, changes in 3D viewpoint, and unpredictable speed variations, resulting in a large number of model parameters and high training difficulty.
An initial gait recognition model is adopted, including a preprocessing network, a spatial feature extraction network, a temporal feature extraction network, and a feature remapping network. Two-dimensional convolutional modules are used to extract features from the training sequence. The training process is optimized by adjusting the model parameters, thereby reducing the number of model parameters and improving the training effect.
While reducing the difficulty of model training, the model is ensured to have the ability to extract temporal features, thus improving the recognition accuracy.
Smart Images
Figure CN115641643B_ABST