Problems with visualizing results
I tested the DIS5K dataset with train.py but the visualization is out of my expectation, can anyone help me please?
this is usually related to the wrongly loaded checkpoint
Hello. Have you resolved the issue? I am also experiencing the same problem.
Hello. Have you resolved the issue? I am also experiencing the same problem. I think caused by the prompts not fitting, when I calculate the bounding boxes as an input through masks or give the exact point prompts, it's fine
Yes I have the same issue. Do you think we could introduce some safety check in the bbox prompt?
I think fixing bbox it could be enough as the current eval is running only on bboxes:
https://github.com/SysCV/sam-hq/blob/322488826bda616798901c6280d13a9a90444ae7/train/train.py#L545