Pose-Based Human Activity Recognition in Clinical Settings

Created using ChatSlide

This project provides an introductory overview and in-depth analysis of Pose-Based Human Action Recognition (HAR). It covers key research objectives, employing skeleton-based pose estimation and BiLSTM for sequence classification. The methodology includes detailed preprocessing and post-processing phases. The discussion extends to dataset use, integrating both public and custom data for superior activity classification results. Challenges like occlusion and multi-person tracking are...