This model is ported from PaddleHub using this script from OpenCV.
Note:
human_segmentation_pphumanseg_2023mar_int8bq.onnxrepresents the block-quantized version in int8 precision and is generated using block_quantize.py withblock_size=64.
Run the following command to try the demo:
# detect on camera input
python demo.py
# detect on an image
python demo.py --input /path/to/image -v
# get help regarding various parameters
python demo.py --helpInstall latest OpenCV and CMake >= 3.24.0 to get started with:
# A typical and default installation path of OpenCV is /usr/local
cmake -B build -D OPENCV_INSTALLATION_PATH=/path/to/opencv/installation .
cmake --build build
# detect on camera input
./build/opencv_zoo_human_segmentation
# detect on an image
./build/opencv_zoo_human_segmentation -i=/path/to/image
# get help messages
./build/opencv_zoo_human_segmentation -h
Results of accuracy evaluation with tools/eval.
| Models | Accuracy | mIoU |
|---|---|---|
| PPHumanSeg | 0.9656 | 0.9164 |
| PPHumanSeg block | 0.9655 | 0.9162 |
| PPHumanSeg quant | 0.7285 | 0.3642 |
*: 'quant' stands for 'quantized'. **: 'block' stands for 'blockwise quantized'.
All files in this directory are licensed under Apache 2.0 License.