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
Describe the bug
When training a new model type, particularly train_detector_mmdet_convnext_cascade.conf on latest VIAME (v0.20.1), if the output folder does not contain a .zip file, then the pipelines do not show up in the desktop version under the 'Trained' dropdown, despite being successfully trained. Web version untested. The model produces a .py. .pipe, .pth, and .txt files but no .zip on success.
For now, I've had them produce a dummy .zip file so that they show up under the 'Trained' dropdown in Desktop, though the .zip file is just an empty file
I think I'm worried that the same situation might be encountered for other model types, e.g. YOLOs, SVMs, MMDet models which also don't produce .zips, but also untested
The text was updated successfully, but these errors were encountered:
mattdawkins
changed the title
New trained model type requiers .zip in output folder even though no .zip required for model
New trained model type requires .zip in output folder even though no .zip required for model
Jul 6, 2023
Describe the bug
When training a new model type, particularly train_detector_mmdet_convnext_cascade.conf on latest VIAME (v0.20.1), if the output folder does not contain a .zip file, then the pipelines do not show up in the desktop version under the 'Trained' dropdown, despite being successfully trained. Web version untested. The model produces a .py. .pipe, .pth, and .txt files but no .zip on success.
For now, I've had them produce a dummy .zip file so that they show up under the 'Trained' dropdown in Desktop, though the .zip file is just an empty file
I think I'm worried that the same situation might be encountered for other model types, e.g. YOLOs, SVMs, MMDet models which also don't produce .zips, but also untested
The text was updated successfully, but these errors were encountered: