205-vision-background-removal

Background Removal Demo

This demo notebook shows image segmentation and removing/adding background with U^2-Net and OpenVINO™.

Notebook Contents

• Importing Pytorch library and loading U^2-Net model.
• Converting PyTorch U^2-Net model to ONNX and OpenVINO IR.
• Loading and preprocessing input image.
• Doing inference on OpenVINO IR model.
• Visualizing results.

U^2-Net source

@InProceedings{Qin_2020_PR,
title = {U2-Net: Going Deeper with Nested U-Structure for Salient Object Detection},
author = {Qin, Xuebin and Zhang, Zichen and Huang, Chenyang and Dehghan, Masood and Zaiane, Osmar and Jagersand, Martin},
journal = {Pattern Recognition},
volume = {106},
pages = {107404},
year = {2020}
}

Installation Instructions

If you have not installed all required dependencies, follow the Installation Guide.