8000 Commits · joestrummer82/DeepLabV3-Tensorflow · GitHub

More Web Proxy on the site http://driver.im/

Commits

Commits on Jan 11, 2018

new val mIOU
NanqingD
committed

Commits on Jan 9, 2018

missed last validation sample
NanqingD 8000
committed
better structure representation
NanqingD
committed

Commits on Dec 25, 2017

fix typo
NanqingD
committed

Commits on Dec 24, 2017

fix typo
NanqingD
committed
add training tricks
NanqingD
committed
finish VOC2012 training
NanqingD
committed

Commits on Dec 22, 2017

small inconsistency, no influence to performance
NanqingD
committed

Commits on Dec 21, 2017

upsample logits
NanqingD
committed
fix architecture
NanqingD
committed
Switch to train_val_voc12.py
NanqingD
committed
fix tf.train.string_input_producer() bug
NanqingD
committed
update README
NanqingD
committed
train and val in one script
NanqingD
committed

Commits on Dec 20, 2017

fix summary bug
NanqingD
committed
fix softmax layer bug and add train_val
NanqingD
committed
fix val_voc12.py bug
NanqingD
committed

Commits on Dec 18, 2017

fix args bug
NanqingD
committed
fix bug for block4
NanqingD
committed
better implementation of weight decay
NanqingD
committed

Commits on Dec 17, 2017

update README
NanqingD
committed
make multi-gpu version loss converge
NanqingD
committed

Commits on Dec 16, 2017

fix batch_norm update bug
NanqingD
committed

Commits on Dec 15, 2017

fix data_factory
NanqingD
committed
fix typo on README
NanqingD
committed
evaluation on miou
NanqingD
committed
update performance
NanqingD
committed
split train and val for training
NanqingD
committed
add SGD
NanqingD
committed
multi-gpu on single machine
NanqingD
committed

Commits on Dec 14, 2017

reformat table
NanqingD
committed
add performance on README
NanqingD
committed
merge to TF1.4 for multi-gpu
NanqingD
committed
Merge pull request #4 from NanqingD/tf_1.4
Nanqing
authored
add multi-gpu on single machine cmd line to README
NanqingD
committed

0