Shiyu Huang

If you only train detector on synthetic data, the result will be bad.

You should fellow the training strategy of our paper: https://arxiv.org/abs/1703.06283

Yes, it is.

  You need change the loss function and data_engine.

You should run these commands in the terminal: `pip install image` `pip install sklearn` `pip install scipy` And you should clone this repo: https://github.com/huangshiyu13/image_pylib , and put it along with...

I don't know your dataset's format, so I can't answer your question. But you can download my datasets and figure out my format. I think my datasets' format is easy...

Hello, thank you for your interest in this project. We will clean our training codes and datasets recently. Our new version be released at: https://github.com/TARTRL/Deepfake_Detection Our implementation has no related...

It seems that your version of pyglet is incorrect. According to the official MPE instructions, you should install pyglet==1.5.27: https://github.com/openai/multiagent-particle-envs?tab=readme-ov-file#getting-started

You can pass your predefined neural network to the OpenRL: Code Examples1: https://github.com/OpenRL-Lab/openrl/blob/8797b0999b0bacdac0658eda54c137f0c06e4544/examples/sb3/train_ppo.py#L37 Code Example2: https://github.com/OpenRL-Lab/openrl/blob/8797b0999b0bacdac0658eda54c137f0c06e4544/examples/nlp/train_ppo.py#L25 Document: https://openrl-docs.readthedocs.io/en/latest/quick_start/train_nlp.html

使用这个脚本：https://github.com/TARTRL/Deepfake_Detection/blob/main/scripts/test.sh