Official implementation of lightweight spatial representation enhancement for RTMPose, featuring P4/P5 semantic-detail fusion and Scale-Adaptive Crop Resolution (SACR).
computer-vision pytorch human-pose-estimation pose-estimation feature-fusion real-time-pose-estimation keypoint-detection animal-pose-estimation mmpose whole-body-pose-estimation rtmpose ap10k
-
Updated
Jul 8, 2026 - Python