MLENet: Multi-level efficient network based on single-scale feature extraction for human keypoint estimation

doi:10.1016/j.eswa.2025.130092

MLENet: Multi-level efficient network based on single-scale feature extraction for human keypoint estimation

2025-10-22

0

PRE

AI

DOI：10.1016/j.eswa.2025.130092

Original

Original request for help

Save

Abstract

En 中文

• The proposed multi-level feature extraction strategy enhances accuracy without multi-scale fusion overhead. • The integration of LFE, SFE, and GFE effectively models local, spatial, and global dependencies among keypoints. • Channel and spatial squeeze attention modules further improve feature discrimination in low-level extraction.

MLENet: Multi-level efficient network based on single-scale feature extraction for human keypoint estimation

Abstract

Journal

Organization