Proceedings of the
The Nineteenth International Conference on Computational Intelligence and Security (CIS 2023)
December 1 – 4, 2023, Haikou, China
A Lightweight Pose Estimation Model Based on Coordinate Classification
State Key Laboratory of Networking and Switching Technology, Beijing University of Posts and Telecommunications, China.
ABSTRACT
Recent studies in the field of 2D pose estimation have indeed demonstrated exceptional performance on a variety of benchmarks. However, such studies often use computationally expensive elements or complicated structural models, making their practical application difficult due to the large number of model parameters and high latency. In this paper, we propose a lightweight model based on coordinate classification. We conduct experiments on the MS COCO and MPII datasets. The results show that our model achieves a score of 72.2 AP on the COCO dataset with only 6.728M parameters and 0.734 GFLOPs.
Keywords: Human pose estimation, Coordinate classification, Lightweight.
Download PDF
State Key Laboratory of Networking and Switching Technology, Beijing University of Posts and Telecommunications, China.
ABSTRACT
Recent studies in the field of 2D pose estimation have indeed demonstrated exceptional performance on a variety of benchmarks. However, such studies often use computationally expensive elements or complicated structural models, making their practical application difficult due to the large number of model parameters and high latency. In this paper, we propose a lightweight model based on coordinate classification. We conduct experiments on the MS COCO and MPII datasets. The results show that our model achieves a score of 72.2 AP on the COCO dataset with only 6.728M parameters and 0.734 GFLOPs.
Keywords: Human pose estimation, Coordinate classification, Lightweight.
Download PDF