search

mapooon / SelfBlendedImages

[CVPR 2022 Oral] Detecting Deepfakes with Self-Blended Images https://arxiv.org/abs/2204.08376
Other
185 stars 32 forks source link

The AUC of the CDF has been 0.5 #28

Open hanxiang323 opened 1 year ago

hanxiang323 commented 1 year ago

Dear expert, I would like to know that my test data on the data set CDF is 0.5, and after 100 rounds of training, batch-size is 16. Then the second layer training is changed into 10 rounds, and the test result is still 0.5. Could you give me some advice on this problem? Thank you very much.

微信图片_20230318224942 微信图片_20230318224950 微信图片_20230318224955 微信图片_20230318224959

mapooon commented 1 year ago

It seems that the CelebDF videos were not loaded properly. Place them following data/datasets.md.

hanxiang323 commented 1 year ago

Hello, my data set should be placed in the right position, but the loaded txt file does have only 518 records as shown in the result picture, so I wonder if there is something wrong with the txt file here. image image

mapooon commented 1 year ago

CelebDFv2 has 518 test videos. So I think there is no problem with the txt file. Please check if videos can be loaded by other ways. For example, by executing following command in python3 interpreter: os.path.isfile('data/Celeb-DF-v2/Celeb-synthesis/videos/id17_id2_0003.mp4') If not, there can be something wrong with your directory.

hanxiang323 commented 1 year ago

Dear Doctor, my path name has made a mistake, and it has now been changed. However, when I run it again, there was an error. image

hanxiang323 commented 1 year ago

Hello, Doctor, can you provide your predict_single module, or is there a download address?