README.md

Real-Time High-Resolution Background Matting

Input

Source image

Background image

(Image from https://drive.google.com/drive/folders/16H6Vz3294J-DEzauw06j4IUARRqYGgRD)

Output

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 background_matting_v2.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 background_matting_v2.py --input IMAGE_PATH --savepath SAVE_IMAGE_PATH

This model requires a background image.
You can specify the background image by put the file path after the --bgr_image option.

$ python3 background_matting_v2.py --input IMAGE_PATH --bgr_image BGR_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 background_matting_v2.py --video VIDEO_PATH

You can specify the "model type" by specifying after the --model_type option. The model type is selected from "mobilenetv2", "resnet50".

$ python3 background_matting_v2.py --model_type mobilenetv2

Reference

• Real-Time High-Resolution Background Matting

Framework

Pytorch

Model Format

ONNX opset=11

Netron

mobilenetv2.onnx.prototxt
resnet50.onnx.prototxt