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
Is there a possibility to retrain the network on my own generated synthetic dataset without losing the current performance of the network? I would like to just retrain some certain layers to get even better performance on my own dataset, or is the architecture too special for that?
I would like to avoid to retrained the network from scratch? For generating the synthetic dataset I used SceneNet RGB-D with custom rooms and objects.
The text was updated successfully, but these errors were encountered:
Is there a possibility to retrain the network on my own generated synthetic dataset without losing the current performance of the network? I would like to just retrain some certain layers to get even better performance on my own dataset, or is the architecture too special for that?
I would like to avoid to retrained the network from scratch? For generating the synthetic dataset I used SceneNet RGB-D with custom rooms and objects.
The text was updated successfully, but these errors were encountered: