Concordia University, Decathlon
Hasib Zunair
[Paper
] [Project
] [Demo
] [BibTeX
]
This is official code for our BMVC 2022 paper:
Title of Your Paper
Summarize in 3-5 sentences your project here.
This code requires Python YOUR_PYTHON_VERSION and CUDA YOUR_CUDA_VERSION. Clone the project repository, then create and activate the following conda envrionment.
# clone repo
git clone https://github.com/hasibzunair/mlcv-readme-template
cd peekaboo
# create env
conda update conda
conda env create -f environment.yml
conda activate myenv
Or, you can also create a fresh environment and install the project requirements inside that environment by:
# clone repo
git clone https://github.com/hasibzunair/mlcv-readme-template
cd peekaboo
# create fresh env
conda create -n myenv python=3.8
conda activate myenv
# install reqs
pip install -r requirements.txt
We expect Dataset1 and Dataset2 datasets to have the following structure:
datasets/
|-- Dataset1/
|---- VOC2007/
|------ JPEGImages/
|------ Annotations/
|------ ImageSets/
......
|-- Dataset2/
|---- annotations/
|---- images/
|------ train2014/
|------ val2014/
...
Add any intructions for pre-processing data, to make it ready for training.
python train.py --dataset Dataset1
python train.py --dataset Dataset2
python eval.py --dataset Dataset1
python eval.py --dataset Dataset2
Refer to supplementary materials if any.
We provide pretrained models on GitHub Releases for reproducibility.
Dataset | Backbone | mIoU(%) | Download |
---|---|---|---|
GLaS | LeViT-UNet 384 | 76.06 | download |
Kvasir & CVC-ClinicDB | LeViT-UNet 384 | 84.02 | download |
NYUDv2 | U-Net++ | 39.31 | download |
Add demo details here.
If you use X in your research, please use the following BibTeX entry.
@inproceedings{zunair2022masked,
title={Masked Supervised Learning for Semantic Segmentation},
author={Zunair, Hasib and Hamza, A Ben},
booktitle={Proc. British Machine Vision Conference},
year={2022}
}
Click to view
[July 12, 2022] Added this fancy drop down!
[March 22, 2022] I started making this template for my own reference.
Give credits to codebases you built on!