MesoNet-Pytorch icon indicating copy to clipboard operation
MesoNet-Pytorch copied to clipboard

The Pytorch Implemention of MesoNet

Results 9 MesoNet-Pytorch issues
Sort by recently updated
recently updated
newest added

定义模型时,在两个地方用了同一个bn,这会导致运行model.eval()测试时和训练时的准确率相差甚远。 ![image](https://user-images.githubusercontent.com/46945475/181872173-4d2bbfab-0dfa-4485-b8b8-e34d27dcdb7c.png)

The requiremnets.txt is missing.

作者你好,我目前研一刚开始这个方向研究,在ff++数据集的那篇文章中,训练集是将所有视频的第一帧取出来还是随机选择一帧?关于他的测试结果,他在对所有抽取视频后的一帧训练完成后,测试时是对整个视频测试,然后统计这个视频每帧的分类结果多数的值,作为该视频的预测,还是也是跟训练一样,取一帧测试,主要是疑惑单取一帧测试感觉达不到99.几的准确率。 希望得到你的帮助,非常感谢

Hello,I am a Chinese student who is just beginning to study fakeface detection, i have no idea to extract the face region from video?can you offer some tools to do...

I want to train a model using the FaceForensics++ dataset , but I don't know how to set the label real/fake when training. Please help me.

Hello, I use this implementation for mesoNet with pre-trained weights that you provided. When I test it on real video from FF++ dataset, the accuracy of the model is very...

各位高手们,一般复现这种老代码,他的python和torch版本都很低,Cuda环境下无法安装他所需的版本,一般是怎么解决的,因为Anaconda不能管理不同的Cuda环境,是重新调试代码使它能在版本更高环境下跑通,还是重新将Cuda降到老版本,