AGenDA

This is the official code for our ICCV 2025 paper:

Adapting Vehicle Detectors for Aerial Imagery to Unseen Domains with Weak Supervision
Xiao Fang, Minhyek Jeon, Zheyang Qin, Stanislav Panev, Celso M de Melo, Shuowen Hu, Shayok Chakraborty, Fernando De la Torre

Requirement

# Create virtual environment
conda create -n agenda python=3.9

# Install torch
conda install pytorch==2.3.1 torchvision==0.18.1 torchaudio==2.3.1 pytorch-cuda=12.1 -c pytorch -c nvidia

# Install dependencies
pip install -r requirements.txt

# Install mmengine and mmcv
mim install mmengine
mim install "mmcv>=2.0.0"

# Install mmdetection
git clone https://github.com/open-mmlab/mmdetection.git
cd mmdetection
pip install -v -e .

# Install mmyolo
git clone https://github.com/open-mmlab/mmyolo.git
cd mmyolo
pip install -v -e .

Data preparation

Please follow the instruction here.

Usage

Stage 1: Data generation

Please follow the instruction here.

Stage 2: Data annotation

Please follow the instruction here.

We upload all checkpoints here. For more usage details, please go through each stage.

Citation

Please cite the paper if you use the code and datasets.

@InProceedings{adapting,
    author    = {Fang, Xiao and Jeon, Minhyek and Qin, Zheyang and Panev, Stanislav and De Melo, Celso and Hu, Shuowen and Chakraborty, Shayok and De La Torre, Fernando},
    title     = {{Adapting Vehicle Detectors for Aerial Imagery to Unseen Domains with Weak Supervision}},
    booktitle = {Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV)},
    month     = {October},
    year      = {2025},
    pages     = {8088-8099}
}

Acknowledgement

The code is built on diffusers, DAAM, and AttnDreamBooth, thanks for their amazing work!