aim-uofa/AdelaiDet

training blendmask on custom doesnt give any results

an99990 opened this issue · 7 comments

an99990 commented

So the training works but when i tested with an image i didnt get any segmented region or predictions.
anyone has an idea on why ?

some log when i try to evaluate it :

Some model parameters or buffers are not found in the checkpoint:
basis_module.seg_head.0.weight
basis_module.seg_head.1.{bias, running_mean, running_var, weight}
basis_module.seg_head.3.weight
basis_module.seg_head.4.{bias, running_mean, running_var, weight}
basis_module.seg_head.6.{bias, weight}
an99990 commented

@Yuliang-Liu do you have any idea on why i am not able to load the weight correctly and evaluate the model ?

kimile599 commented

@Yuliang-Liu do you have any idea on why i am not able to load the weight correctly and evaluate the model ?

hi, got a quic question, what's your cuda version and pytorch version. I had trouble with the compatible issues.

engrjav commented

@an99990 i am facing same issue.. were you able to resolve it?

an99990 commented
  • 👍1
engrjav commented

@an99990 but i have trained centernet with same annotations
does this box inst or cond inst need special labelling?

an99990 commented
engrjav commented