AssetDropper: Asset Extraction via Diffusion Models with Reward-Driven Optimization

       

Installation

git clone https://github.com/Lanjiong-Li/AssetDropper.git
cd AssetDropper

conda create -n assetdropper python=3.10 -y
conda activate assetdropper

# Install torch, torchvision based on your machine configuration
pip install torch==2.4.1 torchvision==0.19.1 torchaudio==2.4.1 --index-url https://download.pytorch.org/whl/cu118

# Install other dependencies
pip install -r requirements.txt

Usage

Prepare Input

To help you get started with your own images, you should follow this simple data structure: Put your own image (.jpg or .png) & corresponding mask (.jpg or .png) & caption in the subdirectory of data.

Here is an overview of data structure:

data
├── Caption/
│   └── example.txt
├── Image/
│   └── example.png 
├── Mask/
│   └── example.png 
└── example.txt (type in image names you want to process)

Get Asset from Reference Image & Mask

Run the following command to get asset from the reference image:

python inference.py \
    --pretrained_model_name_or_path "LLanv/AssetDropper" \
    --data_dir "./data" \
    --output_dir "./output" \
    --txt_name "example" \
    --test_batch_size 8 \
    --guidance_scale 2.0 \
    --num_inference_steps 120 \

• --pretrained_model_name_or_path：Path to the pre-trained AssetDropper model checkpoint.
• --data_dir：Path to the directory containing input images & masks.
• --output_dir：Path to the output directory.
• --txt_name：Name of the file that records the image name you want to process.

Or simply run:

bash inference.sh

ToDo List

• Inference code
• Gradio & Hugging Face demo (Coming Soon)
• Dataset (Coming Soon)

Citation

If you find this work useful for your research, please consider citing:

@article{li2025assetdropper,
  title={AssetDropper: Asset Extraction via Diffusion Models with Reward-Driven Optimization},
  author={Li, Lanjiong and Zhao, Guanhua and Zhu, Lingting and Cai, Zeyu and Yu, Lequan and Zhang, Jian and Wang, Zeyu},
  journal={arXiv preprint arXiv:2506.07738},
  year={2025}
}