You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Why is DefaultFormatBundle used in the train pipeline while during the test pipeline ImageToTensor is used?
The DefaultFormatBundle does some more from line 232 onwards regarding the ground truth information. But the main question there for me is why the image is wrapped in a DataContainer in DFB, but not in ITT. And why .contiguous() is used once and the other time it's np.ascontiguousarray(). In both cases the image is stored in numpy-data structs and the storage on the memory should be contiguous for accessing it with the GPU, I suppose. But why use two different methods instead of consistent use (same with the DC)?
reacted with thumbs up emoji reacted with thumbs down emoji reacted with laugh emoji reacted with hooray emoji reacted with confused emoji reacted with heart emoji reacted with rocket emoji reacted with eyes emoji
-
Question regarding the Coco Dataset config for comprehension:
Why is DefaultFormatBundle used in the train pipeline while during the test pipeline ImageToTensor is used?
The DefaultFormatBundle does some more from line 232 onwards regarding the ground truth information. But the main question there for me is why the image is wrapped in a DataContainer in DFB, but not in ITT. And why .contiguous() is used once and the other time it's np.ascontiguousarray(). In both cases the image is stored in numpy-data structs and the storage on the memory should be contiguous for accessing it with the GPU, I suppose. But why use two different methods instead of consistent use (same with the DC)?
Beta Was this translation helpful? Give feedback.
All reactions