osmr / hpeseginf

Inference script for simultaneous human pose estimation and segmentation models

Simultaneous human pose estimation and segmentation

Inference Python script for simultaneous human pose estimation and segmentation MXNet models.

Installation

pip3 install -r requirements.txt

Download model files form the repo releases.

Usage

python3 demo_mx_hpeseg.py --in-video=<path_to_your_video> --out-video=<path_to_resulting_video> --model=<path_to_model_stem> 
python3 demo_tlite_hpeseg.py --in-video=<path_to_your_video> --out-video=<path_to_resulting_video> --model=<path_to_tfl_model>
python3 demo_onnx_hpeseg.py --in-video=<path_to_your_video> --out-video=<path_to_resulting_video> --model=<path_to_onnx_model>

Example:

python3 demo_mx_hpeseg.py --in-video=../hpeseginf_data/vid2.mp4 --out-video=../hpeseginf_data/hpeseg_result.mp4 --model=../hpeseginf_data/ducnet_mobilenet_w1_coco