GithubHelp home page GithubHelp logo

yupeiyu98 / deep-region-competition Goto Github PK

View Code? Open in Web Editor NEW
40.0 2.0 3.0 2.13 MB

[NeurIPS 2021] Unsupervised Foreground Extraction via Deep Region Competition

Python 100.00%
unsupervised-learning segmentation neurips-2021 pytorch energy-based-model

deep-region-competition's Introduction

Unsupervised Foreground Extraction via Deep Region Competition

teaser

[Paper] [Code]

The official code repository for NeurIPS 2021 paper "Unsupervised Foreground Extraction via Deep Region Competition".

Installation

The implementation depends on the following commonly used packages, all of which can be installed via conda.

Package Version
PyTorch โ‰ฅ 1.8.1
numpy not specified (we used 1.20.0)
opencv-python 4.5.1.48
pandas 1.2.3

Datasets and Pretrained Models

Datasets and pretrained models are available at: https://drive.google.com/drive/folders/1qItekRJcOYBIcVi4ChrcyzwFVl-lrw23?usp=sharing

Please follow the following commands to obtain the CLEVR6 dataset:

# Download `clevr_with_masks_train.tfrecords` from deepmind gcloud
cd drc_workspace/scripts
wget https://storage.googleapis.com/multi-object-datasets/clevr_with_masks/clevr_with_masks_train.tfrecords
python load_clevr_with_masks.py

This will save the generated dataset in the meta folder.

Training

# Train a foreground extractor with specified checkpoint folder
python main.py --checkpoints <TO_BE_SPECIFIED>

You may specify the value of arguments during training. Please find the available arguments in the config.yml.example file in drc_workspace folder. Note that config.yml.example file provides the training parameters on full CUB dataset. Parameters on other datasets and data splits can be found in the drc_workspace/config_gallery folder.

Note that DATA indicates the dataset to use (CUB, DOG, CAR, CLEVR and TEXTURED). The path to your dataset folder, i.e., ROOT_DIR, needs to be specified before running the script.

Testing

# Evaluate the extractor
python test.py --checkpoints <TO_BE_SPECIFIED>

Citation

@inproceedings{yu2021unsupervised,
  author = {Yu, Peiyu and Xie, Sirui and Ma, Xiaojian and Zhu, Yixin and Wu, Ying Nian and Zhu, Song-Chun},
  title = {Unsupervised Foreground Extraction via Deep Region Competition},
  booktitle = {Advances in Neural Information Processing Systems (NeurIPS)},
  month = {December},
  year = {2021}
}

deep-region-competition's People

Contributors

yupeiyu98 avatar

Stargazers

 avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar

Watchers

 avatar  avatar

deep-region-competition's Issues

error in execute python test.py

TypeError: load() missing 1 required positional argument: 'Loader' occurs when I try to execute the provided command for testing

About original images of dog dataset

Hello, I found in car dataset there are original, cropped and resized images, but in Dog dataset it only has resized images. Could you share the original and cropped images of dog dataset? It would help a lot for data preprocessing! Thanks!

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    ๐Ÿ–– Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. ๐Ÿ“Š๐Ÿ“ˆ๐ŸŽ‰

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google โค๏ธ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.