Welcome to the official repository for our research on Refined Generalized Focal Loss (REG) for road asset detection and segmentation on Thai highways. This novel framework leverages advanced mathematical formulations to enhance the detection and segmentation of critical road elements using state-of-the-art vision-based models.
This paper introduces an advanced REG formulation designed to tackle class imbalance and localization challenges in road asset detection. The REG model integrates into vision-based detection and segmentation frameworks to improve accuracy and robustness, especially in complex environments with varying lighting conditions and cluttered backgrounds.
Key Contributions:
For a detailed explanation of the mathematical model and background, please check out our previous work at Refined Generalized Focal Loss Explained.
Our rigorous experiments demonstrate the effectiveness of REG in improving road asset detection and segmentation accuracy. The model’s performance outperforms conventional methods, making it a robust solution for real-world applications.
Results Summary:
To get started, clone the repository and install the required dependencies:
git clone https://github.com/kaopanboonyuen/REG.git
cd REG
pip install -r requirements.txt
Detailed usage instructions and example code can be found in the docs directory. For questions or contributions, please refer to the contributing guidelines.
If you use this work in your research, please cite our paper:
@article{panboonyuen2024REG,
title={REG: Refined Generalized Focal Loss for Road Asset Detection on Thai Highways Using Vision-Based Detection and Segmentation Models},
author={Teerapong Panboonyuen},
year={2024},
url={https://github.com/kaopanboonyuen/REG}
}
For further inquiries, reach out to:
Thank you for visiting our repository! We hope you find our work useful in advancing road asset detection and segmentation.