question about data preprocess code

[Yubo-Shankui]

When I use the provided Data pre-processing code to process my custom data, and I'm sure the quality of the custom data is high and the faces are stable without large motion, the number of track directories generated is far less than the number of videos

[SHYuanBest]

Sorry for the long wait. I think this might be related to the detection accuracy of YOLO, maybe you could try replacing it with https://github.com/deepinsight/insightface. And we will be update the data processing workflow next month. Stay tuned!

[SHYuanBest]

@WuNvRenQiaoDaiMaZiRanShen Introducing ​​OpenS2V-Nexus​​, which consists of: (i) ​​OpenS2V-Eval​​, a fine-grained benchmark, and (ii) ​​OpenS2V-5M​​, a million-scale dataset. Welcome to try it!