MaLP: Manipulation Localization Using a Proactive Scheme

Official Pytorch implementation of CVPR 2023 paper "MaLP: Manipulation Localization Using a Proactive Scheme ".

Vishal Asnani, Xi Yin, Tal Hassner, Xiaoming Liu

Paper Supplementary

Prerequisites

• PyTorch 1.5.0
• Numpy 1.14.2
• Scikit-learn 0.22.2

Getting Started

Datasets

• Every GM is used with different datasets they are trained on. The GM-dataset information is given in Tab. 2 of the supplmentary. Please refer to the test images released by Proactive detection work.
• For new datasets used, please go here.
• The training data is used as CELEBA.

Pre-trained model

The pre-trained model trained on STGAN can be downloaded using the information below:

Model	Link
Localization only	Model
Localization + Detection	Model

Training

• Install VIT transformer package following the instructions in https://github.com/lucidrains/vit-pytorch
• Download the STGAN repository files and pre-trained model from https://github.com/csmliu/STGAN and place the train_loc_det.py file in that folder
• Run the code as shown below:

python train_loc_det.py --data_train "YOUR DATA PATH" --resume --model_path "MODEL PATH" 

For training only localization module, run the code as shown below:

python train_loc.py --data_train "YOUR DATA PATH" --resume --model_path "MODEL PATH" 

Testing using pre-trained models

• Download the pre-trained model using the above links.
• Provide the model path in the code
• Run the code as shown below:

python evaluation_loc_det.py --data_train "YOUR DATA PATH" --resume --model_path "MODEL PATH" 

Visualization

STGAN

STGAN

DRIT

GauGAN

If you would like to use our work, please cite:

@inproceedings{asnani2023pro_loc
      title={MaLP: Manipulation Localization Using a Proactive Scheme}, 
      author={Asnani, Vishal and Yin, Xi and Hassner, Tal and Liu, Xiaoming},
      booktitle={IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
      year={2023}
      
}