README.md

MoveNet

Input

(Image from https://pixabay.com/ja/photos/%E5%A5%B3%E3%81%AE%E5%AD%90-%E7%BE%8E%E3%81%97%E3%81%84-%E8%8B%A5%E3%81%84-%E3%83%9B%E3%83%AF%E3%82%A4%E3%83%88-5204299/)

Model variant: Thunder
Ailia input shape : (1, 256, 256, 3)
Range : [0, 1.0]

Model variant: Lightning
Ailia input shape : (1, 192, 192, 3)
Range : [0, 1.0]

Output

• 2D Keypoint + Confidence : (1, 1, 17, 3)
• Range : [0, 1.0]

Usage

Automatically downloads the onnx and prototxt files on the first run. It is necessary to be connected to the Internet while downloading.

For the sample image,

$ python3 movenet.py

If you want to specify the input image, put the image path after the --input option.
You can use --savepath option to change the name of the output file to save.

$ python3 movenet.py --input IMAGE_PATH --savepath SAVE_IMAGE_PATH

If you want to specify the model variant, put the model variant after the --model_variant option.
You can only choose variants from 'thunder','lightning'. default is 'thunder'

$ python3 movenet.py --input IMAGE_PATH --savepath SAVE_IMAGE_PATH --model_variant lightning

By adding the --video option, you can input the video.
If you pass 0 as an argument to VIDEO_PATH, you can use the webcam input instead of the video file.

$ python3 movenet.py --video VIDEO_PATH

Reference

Code repo for movenet

Framework

TensorFlow

Model Format

ONNX opset = 11

Netron

movenet_thunder.onnx.prototxt

movenet_lightning.onnx.prototxt