3DPoseEstimation

Requirements

pip install opencv-python matplotlib

Download Models

See Release https://github.com/Ascend-Huawei/3DPoseEstimation/releases/tag/v0

Image to 2D Keypoints Model

# Under root of this repo
wget -nc --no-check-certificate 'https://modelzoo-train-atc.obs.cn-north-4.myhuaweicloud.com/003_Atc_Models/AE/ATC%20Model/body_pose_picture/OpenPose_light.pb' -O model/OpenPose_light.pb

# Convert
atc --framework=3 --model=model/OpenPose_light.pb --input_shape="input001:1,368,368,3" --input_format=NHWC --output=model/OpenPose_light --output_type=FP32 --out_nodes="light_openpose/stage_1/ArgMax:0" --soc_version=Ascend310

2D to 3D Lifting Model (VideoPose3D)

# See Release
wget -nc --no-check-certificate https://github.com/Ascend-Huawei/3DPoseEstimation/releases/download/v0/video_pose_3d.om -O model/video_pose_3d.om

Sample Run

python run.py \
    --model2D ./model/OpenPose_light.om \
    --model3D ./model/video_pose_3d.om \
    --input ./data/right_turn.mp4 \
    --output_dir='./outputs'