SpecSeg Network for Specular Highlight Detection and Segmentation in Real-World Images

Official repository for the article @ MDPI Sensors »

Citation

@Article{s22176552,
AUTHOR         = {Anwer, Atif and Ainouz, Samia and Saad, Mohamad Naufal Mohamad and Ali, Syed Saad Azhar and Meriaudeau, Fabrice},
TITLE          = {SpecSeg Network for Specular Highlight Detection and Segmentation in Real-World Images},
JOURNAL        = {Sensors},
VOLUME         = {22},
YEAR           = {2022},
NUMBER         = {17},
ARTICLE-NUMBER = {6552},
URL            = {https://www.mdpi.com/1424-8220/22/17/6552},
ISSN           = {1424-8220},
DOI            = {10.3390/s22176552}
}

The Article PDF is open access and can be downloaded directly from MDPI.

(back to top)

---

Update (01/2024):

There is now a Live Demo of the SpecSeg network on Huggingface. Please feel free to test at Huggingface SpecSeg space. (Note: Doesnt look correct on Safari browser due to Huggingface rendering, use Firefox/Chromimum based browsers please).

Introduction

This repository is the implementation of our paper 'SpecSeg Network for Specular Highlight Detection and Segmentation in Real-World Images'. The developed network and pretrained weights can be used for network training and testing. Please cite the paper if you use them and find them useful.

Full article @ MDPI Sensors

Requirements

• Tested with Python 3.7, Tensorflow 2.5+ on local GPU (CUDA 11.3)
• Tested with Python 3.7, Tensorflow 2.8+ on Google Colab
• Uses the Segmentation Models library

Datasets

The datasets used in the paper can be found at the following Github Repos (property of the respective authors):

1. PSD Dataset, Wu et al.
2. SIHQ Dataset, Fu et al.

Usage

For training on your dataset, please change the path to the dataset root folder in the Jupyter notebook or python script.

(back to top)

To Do:

• Jupyter Notebook (for Colab)
• Pretrained weights
• Python Script

(back to top)

License

Distributed under the GNU General Public License v3.0 License, for research training and/or testing purposes. Pleas cite our paper if you use this code or network. For more details, please see Choose a license and the Liscence.txt file.

(back to top)