training config on COCO + LVIS

SamsungLabs / fbrs_interactive_segmentation

[CVPR2020] f-BRS: Rethinking Backpropagating Refinement for Interactive Segmentation https://arxiv.org/abs/2001.10331

Mozilla Public License 2.0

583 stars 94 forks source link

training config on COCO + LVIS #19

Closed XavierCHEN34 closed 4 years ago

XavierCHEN34 commented 4 years ago

Hello, could you please release your training config for COCO + LVIS, I trained on coco+lvis but got much worse performance than yours. I directly concated the two dataset with torch.utils.data.ConcatDataset. And trained the model for 12 epoches, the @80 @85 is even worse than the model trained on SBD.

ptrvilya commented 4 years ago

Hi, thanks for your interest! Unfortunately we can't share the details now.

brianlai98 commented 3 years ago

@ptrvilya Would it be okay to share the details now?

ptrvilya commented 3 years ago

@brianlai98 HI! Checkout our new paper here https://github.com/saic-vul/ritm_interactive_segmentation . We propose the model that is free of test-time optimization techniques and sets SotA results. The new model is trained on COCO+LVIS.