This is the official repository of "High-Resolution Image Harmonization with Adaptive-Interval Color Transformation".

• Paper

To train our model, a GPU with CUDA support is required.

We built the code using Python 3.9 on Linux with NVIDIA GPUs and CUDA 11.6. The required packages can be installed using the requirements.txt file.

pip install -r requirements.txt

We utilize the iHarmony4 dataset for both training and testing. To use the dataset, the directory path must be updated in config.yaml and config_test_FR.yml. Since some images in the dataset have extremely high resolutions, we resize the HAdobe5k subset so that its largest dimension does not exceed 2048 pixels using ./notebooks/resize_dataset.

To start training, simply run the shell file.

runs/train_AICT.sh

Our pretrained models are available in weights. To evaluate our model, simply set pretrain_path in runs/test_AICT.sh and execute the following command:

runs/test_AICT.sh

If this work is helpful for your research, please consider citing the following BibTeX entry.

@InProceedings{meng2024high,
    author    = {Meng, Quanling and Qinglin, Liu and Li, Zonglin and Lan, Xiangyuan and Zhang, Shengping and Nie, Liqiang},
    title     = {High-Resolution Image Harmonization with Adaptive-Interval Color Transformation},
    booktitle = {Advances in Neural Information Processing Systems (NeurIPS)},
    volume    = {37},
    year      = {2024},
    pages     = {13769--13793}
}