- ailia input shape: (1, 3, 128, 128)
- Range:
- ailia input shape: (1, 3, 256, 256)
- ailia Predict API output:
- ailia Predict API output:
- Segmentation
- Shape: (1, 1, 128, 128)
- Landmarks
- Shape: (1, 31, 4)
- 25 upper-body keypoints (x, y, z, visibility) + 2 virtual keypoints (x, y, z, visibility) + 4 ??? face bounding boxes (x, y, z, visibility)?
- Flag
- Shape: (1,)
- Segmentation
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 blazepose.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 blazepose.py --input IMAGE_PATH --savepath SAVE_IMAGE_PATH
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 blazepose.py --video VIDEO_PATH --savepath SAVE_VIDEO_PATH
Squeeze function in Pytorch model removed as it produces an IF layer in ONNX unsupported in ailia (segmentation output has thus 1 more dim)
PyTorch 1.7.1
ONNX opset = 11