vincent-zhq/mrdf

Code for Cross-Modality and Within-Modality Regularization for Audio-Visual DeepFake Detection

Hi there 👋

mrdf's Issues

Environmental issues

Hello, when I was installing the environment, I found that many packages in the requirements.txt conflict with the environment, are the versions of the packages in there correct?

What is the content of train_1.txt file

dataset

I'm unclear about your process for dividing the dataset; could you possibly provide the processed dataset?

The pretrained weights is of models is on Food101 or FakeAVCeleb?

For direct inference on Food101, we provide the pretrained checkpoint example here. The related dataset splits are also available

Is this a typo? @Vincent-ZHQ

Code problems in training

When I run the train.py file in the virtual environment of the server,the code does not run successfully.The following problems occurred:

The amount of GPU usage during training

Hello, I'm interested in your work. Can you tell me on which type of graphics card the training was conducted and how much GPU usage there is for each card?

The performance of the model on the entire dataset

Hello, I notice that only 10% of the whole dataset is used in the code, does it mean that acc is also based on this part of the data?
And I want to test the model on the whole dataset, but there are some problems happened while loading the ckpt model: Missing key(s) in state_dict: "video classifier.0.weight", "audio classifier.0.weight". Was that model structure conflicts in the code and ckpt model?