Closed acgtyrant closed 6 years ago
Currently MxNet is using static graph to build the network, GPU usage memories are pre-allocated. That's why they don't support disagreement between training shape and validation shape. Thus we are using cropped validations. BTW, this validation is just some evaluation numbers you monitor at training process not what you do for REAL validation benchmark. If you want to get real benchmarks please use the test script to generate results and use https://github.com/mcordts/cityscapesScripts to do evaluation.
The
val_args
copys fromtrain_args
while it does not changecrop
.However, the
data_grep/get_cityscapes_list.py
offersis_crop
, I thinkval_bigger_patch.lst
should not be cropped version. So I setis_crop
asFalse
to produceval_bigger_patch.lst
, and I tried to disable 'crop' intrain/solver.py
as below:But
module.fit
fails and it seems that it complains train_data and val_data is not consistent while their data and label's shape are not same, module is bind to the train_data's shape as below already:If you use cropped val_bigger_patch.lst actually, then I tried to validate it on full image by myself, or the program may be buggy in validating, it not enouge to fix #16 .