Skip to content

My own Gluon reimplement of "A simple yet effective baseline for 3d human pose estimation"

License

Notifications You must be signed in to change notification settings

lck1201/simple-effective-3Dpose-baseline

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

14 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

A simple yet effective baseline for 3D human pose estimation

My own Gluon reimplement of A simple yet effective baseline for 3D human pose estimation
Here is the original implementation

Todo:

  • Provide trained model
  • Provide results on 2D pose estimates as input

Enviroments

python 3.7
mxnet-cu90 1.4.0
CUDA 9.0

Dependency

pip install pyyaml
pip install scipy
pip install matplotlib
pip install easydict

Dataset

  1. Baidu Disk (code: kfsm) or Google Drive to download the HM3.6M annotation
  2. Unzip data under data folder, and organize like this
${PROJECT_ROOT}
    `--data
        `--annot
            `--s_01_act_02_subact_01_ca_01
            `--s_01_act_02_subact_01_ca_02
            `-- ......
            `-- ......
            `-- ......
            `--s_11_act_16_subact_02_ca_04            

How-to-use

You can download my trained model from Google Drive, which MPJPE is 44.9mm.

usage: train.py/test.py [-h] --gpu GPU --root ROOT --dataset DATASET [--model MODEL]
                        [--debug DEBUG]

optional arguments:
  -h, --help         show this help message and exit
  --gpu GPU          GPUs to use, e.g. 0,1,2,3
  --root ROOT        /path/to/project/root/
  --dataset DATASET  /path/to/your/dataset/root/
  --model MODEL      /path/to/your/model/, to specify only when test
  --debug DEBUG      debug mode

Train: python train.py --root /project-root --gpu /gpu-to-use

Test: python test.py --root /project-root --gpu /gpu-to-use --model /model-path

PS: You can modify default configurations in config.py. Because it's a quite simple system, not many hyperparameters need to be tuned.

Results

Since I don't have 2D pose estimate results on HM3.6M, I just experiment with 2D ground truth as input. My best result is 44.9mm(no augment is used), slightly better than 45.5mm reported by paper.

Method Avg Direct Discuss Eating Greet Phone Photo Pose Purch Sitting SittingD Smoke Wait WaitD Walk WalkT
My Result 44.9 36.8 43.5 40.5 43.0 46.2 54.7 40.0 43.6 52.9 59.7 44.2 44.5 45.0 34.6 37.3
Paper 45.5 37.7 44.4 40.3 42.1 48.2 54.9 44.4 42.1 54.6 58.0 45.1 46.4 47.6 36.4 40.4

Figure1

Figure2

Figure3

About

My own Gluon reimplement of "A simple yet effective baseline for 3d human pose estimation"

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages