samsonyubaijian / actionet Goto Github PK

Hi,

Thanks for open-sourcing the dataset! After going through it, I have two questions regarding the hierarchical task annotation:
(1) Is the task description divided into several sub-goals (as shown in Figure 2 in the paper), or just plain text as in the repository?
(2) Do you annotate the alignment of phrases in task descriptions to the video frames as shown in Figure 1 in the paper?

Thanks!
Yichi