Attention this repository is no longer maintained

Please refer to https://github.com/VRI-UFPR/ANDA from now on.

ANDA: A Novel Data Augmentation Technique Applied to Salient Object Detection

Official code for the ICAR 2019 paper "ANDA: A Novel Data Augmentation Technique Applied to Salient Object Detection"

REQUIREMENTS:

We recommend the use of conda alternatively miniconda for python environment management. For the scripts, refer to the requirements.txt in the root folder, for the PConvInpainting refer to PConvInpainting/requirements.txt.

STEP BY STEP USAGE:

• Download MSRA10K at https://mmcheng.net/msra10k/
• Run scripts/protocol.sh PATH ; e.g. MSRA10K_Imgs_GT/Imgs
• Run PConvInpainting/inpaintMSRA10K.py ; --help for parameter instructions
• Run scripts/featureRelated/computeKnn.py ; --help for parameter instructions
• Run scripts/featureRelated/anda.py ; --help for parameter instructions

A bash script for the entire process is available at scripts/run.sh

cd scripts
bash run.sh

CITATION:

If you found this code useful for your research, please cite:

@INPROCEEDINGS{ruiz2019anda,
author={D. V. {Ruiz} and B. A. {Krinski} and E. {Todt}},
booktitle={2019 19th International Conference on Advanced Robotics (ICAR)},
title={ANDA: A Novel Data Augmentation Technique Applied to Salient Object Detection},
year={2019},
pages={487-492},
keywords={feature extraction;learning (artificial intelligence);neural nets;object detection;video signal processing;image cropping;ANDA technique;labeled salient objects;image inpainting;background information;data augmentation technique;salient object detection context;mean absolute error;F-measure},
doi={10.1109/ICAR46387.2019.8981655},
ISSN={null},
month={Dec},}

DISCLAIMER:

• This is a research code, so compatibility issues might happen.
• The PConvInpainting folder contain code from the REPOSITORY.