MZ2FY
MZ2FY
honggu,您好。我最近在利用Xception训练deefake,其中我遇到了一些问题:我的训练精度非常高,但是validation和test的acc却很低或者不变。起初我以为是我Dataloader部分的代码写错了,但是我将train dataset作为validation,却能够在每个epoch下acc能够提升。先声明一下,我采用的不是FF++的数据集和Kaggle上DFDC的full数据集(太大了),而是用的Kaggle上给的sample dataset(大概400个训练视频,400个测试视频),并且在提取人脸后也做了样本平衡的操作。所以,我想问一下经验丰富的您,是否是我采用的数据集太小而导致的问题,是否必须采用full dataset才能够在validation和test中看到一些效果?
Hi, This is a good work. and I didn't see any command or options in the inference code for editing attributes. DO I miss something? Could you explain it? Thanks.
Hi, can you provide a guide to use the code? Thanks!
Hi, Xiaoyi, this is a remarkable work. Here, I want to ask about some question: **1. In table 1 of the paper, are the results of FFD and CNNDetect obtained...