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
Thank you for this great work ! I'm wondering if it can be used to efficiently load sequences of images, for example for video prediction. Imagine you have a large dataset of images from cameras (ordered images, eg. one folder per day) and you want to sample a batch of B samples, each sample being a sequence of p+f consecutive images (p input frames, f future frames to predict). Is there an efficient way to do this ?
The text was updated successfully, but these errors were encountered:
@volcverse You just need to write a custom TraversalOrder that associates sequences of images together in each minibatch. #374 allows feeding any custom data you need to do this into your custom TraversalOrder class
Thank you for this great work ! I'm wondering if it can be used to efficiently load sequences of images, for example for video prediction. Imagine you have a large dataset of images from cameras (ordered images, eg. one folder per day) and you want to sample a batch of B samples, each sample being a sequence of p+f consecutive images (p input frames, f future frames to predict). Is there an efficient way to do this ?
The text was updated successfully, but these errors were encountered: