A Trend of 2D Human Pose Estimation Base on Deep Learning
Chunyan Wei, Wei Long, Sudong Jiang, Changxue Chen, Dongfang Wu, Linhua Jiang · 2022
The research on Human pose estimation remains the most fundamental and challenging problem in computer vision. In this context, computer vision-based automobile safety-assisted driving technology has received comprehensive attention and is widely used in automobile-assisted driving systems. Powered by advanced artificial intelligence, the scale of data to be learned by deep learning networks is increasing, the processing tasks are becoming more and more complex, and the model parameters are also increasing. In the field of distributed parallel computing research has also received extensive attention. In this work, we focus on the challenge of human posture estimation in two dimensions, using both conventional and deep learning techniques, with an emphasis on its study's potential use in traffic-related settings. Recent approaches for estimating human poses are introduced, and a summary of how to extract bone joint points using deep learning techniques is provided. In conclusion, it elaborates on the present state of going beyond the bounds of single-computer computing resources and merging distributed architecture with deep learning to address the challenges and opportunities in human pose estimation. Human posture estimation has several technological applications, including the detection of driving behaviour and the analysis of fitness data.