Log-M.Y.CHEN

Although great progress has been made in estimating 3D human motion and shape from dynamic observations, it is still a challenge to recover a credible pose sequence in the presence of noise and occlusion. 一种基于新型条件VAE的可生成3D人体运动模型，可实现表达性和一般的运动重建和生成; 随后的一个robust测试时间优化的方法，使用Model作为一个强大的运动，联合解决姿态、身体形状和地平面/接触; 通过广泛的评估，可以对各种输入（RGB- D视频，2D/3D 关节位置序列）进行操作，从而产生准确可信的运动和接触。

Machine Learning Mini-Project A 3D Human Model for Robust Shape

# Machine Learning Mini-Project A 3D Human Model for Robust Shape Estimation # Environment Setup + If based on Server Ubuntu or Linux: > `conda install pytorch==1.6.0 torchvision==0.7.0 cudatoolkit=10.1 -c pytorch` + For other cases(Mac or Windows): 1. Install **CUDA 10.1** : > use 'pip install' 2. Install **Pytorch 1.6.0**: > use 'pip install' 3. Install libs **matplotlib, scikit-learn** > use 'pip install' 4. **smpl+h**:

Data Preprocessing

Before Data Analysis, we should complete the data preprocessing step in order to clean the data and make the data analysis result more accurate and excellent. This is the demo resource which is used in the BLOG 'Data Preprocessing Demo'. Download the file directly and use it by no matter what programming language including python or java. Enjoy yourself and Become more and more experienced !!!