Paper reproduction: expressive body capture

Write it at the front ：

01, For purely personal learning , Do not do any business . If there is any infringement , Contact me to delete .

02, Limited personal ability , For reference only , Welcome to discuss , Grow up together .

One . Paper information

《Expressive Body Capture: 3D Hands, Face, and Body from a Single Image》

Address of thesis ：https://arxiv.org/pdf/2104.03176.pdf

Code address ：https://github.com/vchoutas/smplify-x

Two . Content abstract

  This is the renderings ,

To tell you the truth, I feel like this paper is a blockhouse , I also want to implement my own data . Then I found a problem ：

The dataset is EHF（Expressive Hands and Face）

EHF（Expressive Hands and Face）

EHF It's used to evaluate 3D Data set of human body reconstruction method （ common 601 File ）.

EHF Data set containing 100 frame , Carefully marked by professional annotators , In order to ensure that 3D Pseudo truth value （pseudo ground truth
） Authenticity .

1.  Each frame contains ：

1) - file name ：“*_img.jpg” - jpg Format RGB Full body image .

2) - file name ：“*_img.png” - png Format RGB Full body image .

3) - file name ：“*_scan.obj” -  In this frame 3D scanning （ Multi view stereo reconstruction ）.

4) - file name ：“*_align.ply” -  With 3D mesh Form will SMPL-X And 3D Scan alignment （ As a pseudo truth value ）.

5) - file name ：“*_2Djnt.json” -  Use OpenPose Estimate of the 2D The joints （ From monocular RGB）.

6) - file name ：“*_2Djnt.png”-OpenPose 2D Visualization of joints .

 scanning （.obj） And alignment （.ply） All are in meters .

Simply speaking ： A picture , It needs to be matched 6 File . 

3、 ... and . Code implementation

The article does not reproduce the popular science , Basic algorithmic knowledge is required .

git clone [email protected]:muelea/smplify-xmc.git
cd smplify-xmc

Download the trained model and test data .

Call me Lei Feng , The network disk link is as follows ：

 link ：https://pan.baidu.com/s/1_5fgni7u7NylZKrLYWOwPA 
 Extraction code ：yq3w 

The code structure is as follows ：

 

 

  If the package and dependency are OK , function demo Inference file

What needs to be noted here is , The consistency between the path and the path in your own code .（ If something goes wrong, compare it with the error report .）

cd /smplify-xmc
python main_singleview.py --config configs/fit_smplx_singleview.yaml \
--dataset mtp_demo \
--input_base_dir data/example_input/singleview/subject1 \
--input_dir_poses data/example_input/presented_poses \
--output_dir data/example_output/singleview/subject1 \
--model_folder /smplify-xmc/modelfolder/models/

Comparison of renderings

 

 

END

Statement : This article is for personal study only , Because of the author's limited ability , Relevant views and information are for reference only .

If there is infringement, please contact .

Welcome readers to ask questions and exchange .

Our goal is the sea of stars ！

（ Take your brother dddd）