Noted: The paper is under review, and this code repository will be updated after it is accepted

RSDet

Removal and Selection: Improving RGB Infrared Object Detection via Coarse to Fine Fusion

Based on the MMdetection 3.1.0 framework, this project modifies its data flow and related classes and functions, and changes the MMdetection to a multi-modal detection framework to facilitate RGBT Object Detection.

Overview

Main Results

Dataset: FLIR, LLVIP and Kaist

Getting Started

Installation

ref : mmdetection installation

Step 1: Clone the RSDet repository:

To get started, first clone the RSDet repository and navigate to the project directory:

git clone https://github.com/Zhao-Tian-yi/RSDet.git
cd RSDet

Step 2: Environment Setup:

RSDet recommends setting up a conda environment and installing dependencies via pip. Use the following commands to set up your environment:

Create and activate a new conda environment

conda create -n RSDet
conda activate RSDet

If you develop and run mmdet directly, install it from source

pip install -v -e .

Install Dependencies

pip install -r requirements.txt
pip install -r requirements_rgbt.txt

Result

Kaist Result

Citation

@article{zhao2024removal,
  title={Removal and Selection: Improving RGB-Infrared Object Detection via Coarse-to-Fine Fusion},
  author={Zhao, Tianyi and Yuan, Maoxun and Wei, Xingxing},
  journal={arXiv preprint arXiv:2401.10731},
  year={2024}
}

✅ Updates

March. 19th, 2024: Update: we have updated the source about one version bug.
March. 12th, 2024: The source code is provided.

Zhao-Tian-yi/RSDet