UACANet: Uncertainty Augmented Context Attention for Polyp Segmentation

Official pytorch implementation of UACANet: Uncertainty Augmented Context Attention for Polyp Segmentation
To appear in the Proceedings of the 29th ACM International Conference on Multimedia (ACMMM 2021)

Abstract

We propose Uncertainty Augmented Context Attention network (UACANet) for polyp segmentation which consider a uncertain area of the saliency map. We construct a modified version of U-Net shape network with additional encoder and decoder and compute a saliency map in each bottom-up stream prediction module and propagate to the next prediction module. In each prediction module, previously predicted saliency map is utilized to compute foreground, background and uncertain area map and we aggregate the feature map with three area maps for each representation. Then we compute the relation between each representation and each pixel in the feature map. We conduct experiments on five popular polyp segmentation benchmarks, Kvasir, CVC-ClinicDB, ETIS, CVC-ColonDB and CVC-300, and achieve state-of-the-art performance. Especially, we achieve 76.6% mean Dice on ETIS dataset which is 13.8% improvement compared to the previous state-of-the-art method.

News 🚀

Our new work on high-resolution salient object detection, Revisiting Image Pyramid Structure for High Resolution Salient Object Detection has been accepted to ACCV2022. Come and see the results!

0. Automatic Installation

• (21.10.08 added) by simply using install.sh, you can download and locate train/test datasets, backbone checkpoints, pretrained checkpoints and pre-computed results, and automatically create conda environment named uacanet. If you find error using our script, please create environment and prepare datasets manually.

1. Create environment

• Create conda environment with following command conda create -n uacanet python=3.7
• Activate environment with following command conda activate uacanet
• Install requirements with following command pip install -r requirements.txt

2. Prepare datasets

• Download Train/Test datasets and Res2Net backbone checkpoint from following URL
• Move folder data to the repository.
• Folder should be ordered as follows,

|-- configs
|-- data
|   |-- TestDataset
|   |   |-- CVC-300
|   |   |   |-- images
|   |   |   `-- masks
|   |   |-- CVC-ClinicDB
|   |   |   |-- images
|   |   |   `-- masks
|   |   |-- CVC-ColonDB
|   |   |   |-- images
|   |   |   `-- masks
|   |   |-- ETIS-LaribPolypDB
|   |   |   |-- images
|   |   |   `-- masks
|   |   `-- Kvasir
|   |       |-- images
|   |       `-- masks
|   `-- TrainDataset
|       |-- images
|       `-- masks
|-- lib
|   |-- backbones
|   |-- losses
|   `-- modules
├── LICENSE
├── README.md
├── requirements.txt
├── results
│   ├── result_CVC-300.csv
│   ├── result_CVC-ClinicDB.csv
│   ├── result_CVC-ColonDB.csv
│   ├── result_ETIS-LaribPolypDB.csv
│   └── result_Kvasir.csv
├── run
│   ├── __init__.py
│   ├── Eval.py
│   ├── Inference.py
│   ├── Test.py
│   └── Train.py
├── snapshots
│   ├── UACANet-L
│   │   ├── CVC-300
│   │   ├── CVC-ClinicDB
│   │   ├── CVC-ColonDB
│   │   ├── ETIS-LaribPolypDB
│   │   ├── Kvasir
│   │   └── latest.pth
│   └── UACANet-S
│       ├── CVC-300
│       ├── CVC-ClinicDB
│       ├── CVC-ColonDB
│       ├── ETIS-LaribPolypDB
│       ├── Kvasir
│       └── latest.pth
└── utils
    ├── custom_transforms.py
    ├── dataloader.py
    ├── eval_functions.py
    ├── __pycache__
    └── utils.py

3. Train & Evaluate

Train

# Single GPU
CUDA_VISIBLE_DEVICES=0 python run/Train.py --config configs/UACANet-L.yaml --verbose --debug

# Multi GPU (e.g., 0 and 1)
CUDA_VISIBLE_DEVICES=0,1 python -m torchrun --nproc_per_node 2 run/Train.py --config configs/UACANet-L.yaml --verbose --debug

Test

# Generate prediction for benchmarks
python run/Test.py --config configs/UACANet-L.yaml --verbose

Evaluate

# Evaluate on various metrics (e.g., S-measure, E-measure, etc.)
python run/Eval.py --config configs/UACANet-L.yaml --verbose

All-in-One command

# Train, Test, and Evaluate with single command

# Single GPU
CUDA_VISIBLE_DEVICES=0 python Expr.py --config configs/UACANet-L.yaml --verbose --debug

# Multi GPU
CUDA_VISIBLE_DEVICES=0,1 python -m torchrun --nproc_per_node 2 Expr.py --config configs/UACANet-L.yaml --verbose --debug

Pre-computed Maps

Download our best result checkpoint and pre-computed maps from following URL for UACANet-L and UACANet-S.

4. Experimental Results

• UACANet-S

dataset              meanDic    meanIoU    wFm     Sm    meanEm    mae    maxEm    maxDic    maxIoU    meanSen    maxSen    meanSpe    maxSpe
-----------------  ---------  ---------  -----  -----  --------  -----  -------  --------  --------  ---------  --------  ---------  --------
CVC-300                0.902      0.837  0.886  0.934     0.974  0.006    0.976     0.906     0.840      0.959     1.000      0.992     0.995
CVC-ClinicDB           0.916      0.870  0.917  0.940     0.965  0.008    0.968     0.919     0.873      0.942     1.000      0.991     0.995
Kvasir                 0.905      0.852  0.897  0.914     0.948  0.026    0.951     0.908     0.855      0.911     1.000      0.976     0.979
CVC-ColonDB            0.783      0.704  0.772  0.848     0.894  0.034    0.897     0.786     0.706      0.801     1.000      0.958     0.962
ETIS-LaribPolypDB      0.694      0.615  0.650  0.815     0.848  0.023    0.851     0.696     0.618      0.833     1.000      0.887     0.891

• UACANet-L

dataset              meanDic    meanIoU    wFm     Sm    meanEm    mae    maxEm    maxDic    maxIoU    meanSen    maxSen    meanSpe    maxSpe
-----------------  ---------  ---------  -----  -----  --------  -----  -------  --------  --------  ---------  --------  ---------  --------
CVC-300                0.910      0.849  0.901  0.937     0.977  0.005    0.980     0.913     0.853      0.940     1.000      0.993     0.997
CVC-ClinicDB           0.926      0.880  0.928  0.943     0.974  0.006    0.976     0.929     0.883      0.943     1.000      0.992     0.996
Kvasir                 0.912      0.859  0.902  0.917     0.955  0.025    0.958     0.915     0.862      0.923     1.000      0.983     0.987
CVC-ColonDB            0.751      0.678  0.746  0.835     0.875  0.039    0.878     0.753     0.680      0.754     1.000      0.953     0.957
ETIS-LaribPolypDB      0.766      0.689  0.740  0.859     0.903  0.012    0.905     0.769     0.691      0.813     1.000      0.932     0.936

• Qualitative Results

5. Citation

@inproceedings{kim2021uacanet,
  title={UACANet: Uncertainty Augmented Context Attention for Polyp Segmentation},
  author={Kim, Taehun and Lee, Hyemin and Kim, Daijin},
  booktitle={Proceedings of the 29th ACM International Conference on Multimedia},
  pages={2167--2175},
  year={2021}
}

• Conference version will be added soon.

6. Acknowledgement

• Basic training strategy, datasets and evaluation methods are brought from PraNet. Especially for the evalutation, we made Python version based on PraNet's MatLab version and verified on various samples. Thanks for the great work!
• This work was supported by Institute of Information & communications Technology Planning & Evaluation(IITP) grant funded by the Korea government(MSIT) (No.B0101-15-0266, Development of High Performance Visual BigData Discovery Platform for Large-Scale Realtime Data Analysis), (No.2017-0-00897, Development of Object Detection and Recognition for Intelligent Vehicles) and (No.2018-0-01290, Development of an Open Dataset and Cognitive Processing Technology for the Recognition of Features Derived From Unstructured Human Motions Used in Self-driving Cars)