Closed hyf015 closed 7 months ago
Hi Yifei!
When annotating goal-step, we realized some of the videos from Ego4D could be concatenated back-to-back to provide longer storyline. This was important for goal-step annotation as having long temporal context is beneficial for identifying goals and steps. Hope this clarifies!
Hi, thank you very much for the release of this dataset! Just a quick question, I wonder what the video IDs that start with "grp-" mean in the annotation files?