Edge-Guided Feature Fusion Network for RGB-T Salient Object Detection
Author Information
Author(s): Chen Yuanlin, Sun Zengbao, Yan Cheng, Zhao Ming
Primary Institution: Shanghai Maritime University
Hypothesis
The proposed Edge-Guided Feature Fusion Network (EGFF-Net) will improve RGB-T salient object detection by effectively integrating cross-modal information and enhancing edge features.
Conclusion
The EGFF-Net outperforms existing methods in RGB-T salient object detection by effectively integrating cross-modal information and refining object boundaries.
Supporting Evidence
- The proposed method achieved superior performance on benchmark datasets compared to state-of-the-art methods.
- EGFF-Net effectively suppresses background noise and enhances salient object boundaries.
- The model demonstrated robustness in various challenging scenarios, including cluttered backgrounds and low-light conditions.
Takeaway
This study created a new way to find important parts of images using both regular and thermal pictures, making it better at spotting things even in messy backgrounds.
Methodology
The study used a double-input end-to-end network structure with cross-modal feature extraction, edge-guided feature fusion, and salience map prediction.
Limitations
The model's computational complexity may hinder real-time applications, and its effectiveness may decrease in scenarios with ambiguous edges or extreme occlusion.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website