Summary of The Solution For the Gaiic2024 Rgb-tir Object Detection Challenge, by Xiangyu Wu et al.
The Solution for the GAIIC2024 RGB-TIR object detection Challenge
by Xiangyu Wu, Jinling Xu, Longfei Huang, Yang Yang
First submitted to arxiv on: 4 Jul 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Machine Learning (cs.LG)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary This report presents a solution for RGB-TIR object detection from the perspective of unmanned aerial vehicles, leveraging both RGB and TIR images for complementary information during detection. The challenges of RGB-TIR object detection include complex image backgrounds, lighting changes, and uncalibrated image pairs. To address these issues, the authors utilize a lightweight YOLOv9 model with extended multi-level auxiliary branches to enhance robustness. A fusion module is also incorporated into the backbone network for feature-level image fusion, implicitly addressing calibration issues. The proposed method achieves an mAP score of 0.516 and 0.543 on benchmarks A and B respectively while maintaining high inference speed. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary RGB-TIR object detection from unmanned aerial vehicles aims to use both RGB and TIR images together. This is tricky because the images can be complex, lighting can change, and the images may not be perfectly matched. To solve this problem, scientists used a special model called YOLOv9 that has extra branches to help it work better. They also added a way to combine the two types of images so they can work together well. This method worked really well, with an accuracy score of 0.516 and 0.543 on different tests. |
Keywords
* Artificial intelligence * Inference * Object detection
