Comparing to Alayrac et al.

[fukka]

Great work. Just wondering how did you use 'The Visual Centrifuge: Model-Free Layered Video Representations' paper in your paper comparison? Did you retrain that on your data? Or did you get their weights?