https://github.com/chandrasekaraditya/ReMOVE
Official Implementation for "ReMOVE: A Reference-free Metric for Object Erasure"
https://github.com/chandrasekaraditya/ReMOVE
Last synced: about 1 month ago
JSON representation
Official Implementation for "ReMOVE: A Reference-free Metric for Object Erasure"
- Host: GitHub
- URL: https://github.com/chandrasekaraditya/ReMOVE
- Owner: chandrasekaraditya
- License: mit
- Created: 2024-04-13T13:08:24.000Z (over 1 year ago)
- Default Branch: main
- Last Pushed: 2024-04-30T05:35:36.000Z (over 1 year ago)
- Last Synced: 2024-12-30T15:36:35.879Z (10 months ago)
- Language: Python
- Homepage:
- Size: 10.7 MB
- Stars: 11
- Watchers: 1
- Forks: 1
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
- awesome-diffusion-categorized - [Code
README
# ReMOVE
This is the official PyTorch implementation for the CVPRW 24 paper: "ReMOVE: A Reference-free Metric for Object Erasure".
## Requirements
This code was tested with `python=3.8.18`, `pytorch=2.0.1` and `torchvision=0.8.2`. Please follow the instructions [here](https://pytorch.org/get-started/locally/) to install PyTorch and TorchVision dependencies. Installing both PyTorch and TorchVision with CUDA support is strongly recommended.
**NOTE**: This codebase requires SAM to be installed. To install SAM, please follow the instructions [here](https://github.com/facebookresearch/segment-anything/blob/main/README.md#installation).
```
pip install git+https://github.com/facebookresearch/segment-anything.git
```
## Getting Started
### Setup ###
Start by downloading the model at `models/url.txt` (~2.4GB). Currently, `main.py` uses the ViT-H model, but other [models](https://github.com/facebookresearch/segment-anything/blob/main/README.md#model-checkpoints) can be used by modifying `L49` of `main.py`.
```
wget -i models/url.txt -P models/
```
### Usage ###
The `main.py` script takes the following arguments
* `--image_path` : Path to the image.
* `--mask_path` : Path to the corresponding mask.
* `--crop` (_optional_) : Crops the image using a bounding box around the object mask.
* `--draw` (_optional_) : Draws a bounding box around the object (saved to `masked.png`) if `--crop` is used.
A set of sample images with objects removed using various methods are provided in `./examples/`.
#### Bad Inpainting
```
python main.py -i 'examples/sdinpaint/823000000001.jpg' -m 'examples/masks/823000000001.jpg'
```
```
python main.py -i 'examples/sdinpaint/823000000001.jpg' -m 'examples/masks/823000000001.jpg' --crop
```
#### Good Inpainting
```
python main.py -i 'examples/sdinpaint/122000000009.jpg' -m 'examples/masks/122000000009.jpg'
```
```
python main.py -i 'examples/sdinpaint/122000000009.jpg' -m 'examples/masks/122000000009.jpg' --crop
```
## Citation
If you use ReMOVE or find this work useful for your research, please use the following BibTeX entry.
```bibtex
@InProceedings{Chandrasekar_2024_CVPR,
author = {Chandrasekar$^*$, Aditya and Chakrabarty$^*$, Goirik and Bardhan, Jai and Hebbalaguppe, Ramya and AP, Prathosh},
title = {ReMOVE: A Reference-free Metric for Object Erasure},
booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops},
month = {June},
year = {2024}
}
```