Code for the paper "Location-aware Single Image Reflection Removal"

Location-aware Single Image Reflection Removal

The shown images are provided by the datasets from IBCLN, ERRNet, SIR² and the Internet images.

The code and pretrained model for our paper: Location-aware Single Image Reflection Removal [Arxiv Preprint]

Prerequisites

Our code has been tested under the following platform and environment:

• Ubuntu. CPU or NVIDIA GPU + CUDA, CuDNN
• Python 3.7.3, Pytorch 1.2.0
• Requirements: numpy, tqdm, Pillow, dominate, scikit-image

Setup

• Clone or Download this repo
• $ cd Location-aware-SIRR
• $ mkdir model
• Download the pretrained model here
• Move the downloaded model(model.pth) to ./model folder

Usage

• The example test images are provided in ./test_images/blend folder
• If you have ground truth blackground images, put them into ./test_images/transmission folder ( Note that the same pair of images need to be named the same ).
• Run python3 inference.py
• The inference results are in the ./results folder

Citation

If you find our work helpful to your research, please cite our paper.

@article{dong2020location,
  author = {Zheng Dong and Ke Xu and Yin Yang and Hujun Bao and Weiwei Xu and Rynson W.H. Lau},
  title = {Location-aware Single Image Reflection Removal},
  journal={ArXiv},
  volume={abs/2012.07131},
  year = {2020}
}