Lightweight modal-guided cross-attention fusion network for visible-infrared object detection

This is an official PyTorch implementation for our LCAFNet. Paper can be download in LCAFNet

1. Dependences

Create a conda virtual environment and activate it.

1. conda create --name MOD python=3.9
2. conda activate MOD
3. pip install -r requirements.txt

2. Datasets download

Download these datasets and create a dataset folder to hold them.

1. FLIR dataset: FLIR
2. LLVIP dataset: LLVIP
3. M3FD dataset: M3FD
4. MFAD dataset: MFAD

3. Pretrained weights

Download our LCAFNet weights and create a weights folder to hold them.

1. FLIR dataset: LCAFNet_FLIR.pt
2. LLVIP dataset: LCAFNet_LLVIP.pt
3. M3FD dataset: LCAFNet_M3FD.pt
4. MFAD dataset: LCAFNet_MFAD.pt

4. Training our LCAFNet

Dataset path, GPU, batch size, etc., need to be modified according to different situations.

python train.py

5. Test our LCAFNet

python test.py

6. Citation

If you find LCAFNet helpful for your research, please consider citing our work.

@article{Wu2026,
  author       = {Wencong Wu and
                  Hongxi Zhang and
                  Xiuwei Zhang and
                  Hanlin Yin and
                  Yanning Zhang},
  title        = {Lightweight modal-guided cross-attention fusion network for visible-infrared object detection},
  journal      = {Pattern Recognition},
  volume       = {177},
  pages        = {113350},
  year         = {2026}
}