Hi @miguelmartin. I am trying to use RGB and gaze data together. It seems to me that the gaze information (gaze 2d file with frame number and x, y coordinates). For some videos with the has_gaze flag, the number of frames does not cover the entire video, at least not with the same fps. So, how can we use the gaze paired with the video data? The benchmark I am working on is the key step recognition, and from my analysis, only a couple of takes do not have the gaze data.
Is it possible that the gaze files are at 10fps compared to 30fps of the RGB frames?
Resolved in another thread: Ego-exo4d Eye Gaze usage