Closed aiwithshekhar closed 4 years ago
Hi, I think that the simplest option is to obtain bounding box from predicted segmentation mask.
Thanks a lot for replying, 1) If i want to train SSD considering bounding box from fbrs segmentation mask as Ground truth and SSD output as prediction and trying to minimize the loss. Will this work? 2) Can i continue with the multipoint sampler for this?
Hi maybe my question is not relevant to this code base.
Can this repo with code modifications used for click based Bounding Box generation.