Pedestrian pose estimation can provide a basis for planning and control systems of campus unmanned delivery vehicles to ensure pedestrian safety and improve delivery efficiency. A lightweight pedestrian pose estimation model is built in this paper, and a model training method incorporating campus pedestrian information is proposed. In order to verify the effectiveness of the proposed method and the real-time performance of the model, a comparative experiment is conducted and the real-time performance is tested on the server side and the AGV platform. The results show that the proposed model and training method can efficiently and accurately accomplish campus pedestrian pose estimation in real scene.
Pedestrian Pose Estimation for Campus Unmanned Delivery Vehicles
2023-10-27
2285841 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
Pedestrian Behavior Interpretation from Pose Estimation
IEEE | 2021
|Pedestrian Motion State Estimation From 2D Pose
IEEE | 2020
|