About the clean version of FF++ dataset you used in Quantitative Comparison.

[NNNNAI]

Thanks for sharing your work. It is really marvelous!

As you mentioned in the "Experiments on FaceForensics++" section: "However, as FaceForensics++ is not designed for face recognition, some videos display repeated identities. For example, videos numbered 043 and 343 show Vladimir Putin, and videos of 179; 183 and 826 show the same person Barack Obama. Thus, we manually categorize all videos into 885 identities.", would you be kind to share this videos identities list, that is, which videos have the same id?

Thanks again~ have a nice day.

[zyainfal]

Hi, We used version 4 of FF++ and ​I find the vid2pids.npy file that records the dictionary of {video id: person id}. You can read it by

# load dictionary
myDicy = np.load('vid2pids.npy', allow_pickle=True).item()

# check items
for k in myDict.keys():
    print(k, myDict[k])

Please note you need to unzip the following file to get the npy file. Download here: vid2pids.zip