Mask_rcnn: Is it possible to finetune coco weights on a dataset that only has bboxes?

Created on 7 Feb 2018  路  3Comments  路  Source: matterport/Mask_RCNN

I want to get rough segmentation for new labels basically.

Most helpful comment

That is a transfer learning problem. This paper may help you
Learning to segment everything. https://arxiv.org/abs/1711.10370

All 3 comments

That is a transfer learning problem. This paper may help you
Learning to segment everything. https://arxiv.org/abs/1711.10370

@JackieZhangdx I was asking if the code in this repo supports it. Sorry if it wasn't clear.
Would it be complicated to adopt this code for such a scenario?

It doesn't support it out of the box, but it shouldn't be too hard to add that if you're willing to dive into the code and make changes.

Was this page helpful?
0 / 5 - 0 ratings