[Feature] When will mmdet support reading the image mask directly instead of COCODATASET on the instance segmentation task?

[330205812]

What's the feature?

E.g., the training data can be organized as follows: --Annotations --0000.xml --0001.xml ... --Images --0000.jpg --0001.jpg ... --Segmentation --0000.png --0001.png ...

Thanks for reading!

Any other context?

No response

[Czm369]

Thanks for the suggestion. But we don't have this plan in the short term.