Answers are grounded in this paper and cite their sources.
Skeleton-based human detection and activity recognition are greatly affected by computer vision and are available in real-world applications such as security, the physical condition of human movement, robotics, smart home systems, and visual reality applications.Currently, several approaches have been established by using 3D skeleton data and it is directly taken from sensor devices.That is cost-effective and constraints on different lighting conditions, distance scales, and apparatus devices in real-life applications.To overcome these conditions, the proposed approach is considered on 2D to 3D skeleton model with deep learning approach to recognize more reliable information between skeleton joint sequences and human activity changes over time.Our method is used with OpenPose detector to extract effective 2D key points of body parts and reconstruct 3D skeleton model to recognize occlusion of body parts and more accurately on human activities.The experiment is tested with HMDB 51 dataset and our daily-life data and gets efficient results as 94% for human activity recognition.
No takes yet. Share an insight, caveat, or question.
Win et al. (2024) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: