Homec4science
Diffusion lineseg (master) Graph

R964/main.py
master

History Graph

CommitAuthorDetailsCommitted
240d691f9f63Sandro Ivo Sebastiano de Zanet
adapted disparity computation with mask 
Feb 7 2017
5eb1bdecb07cdezanet
removed matplotlib 
Feb 6 2017
6c2f431aaaaedezanet
masked loss 
Feb 1 2017
0fa456f74b0adezanet
removed augmentation 
Jan 30 2017
c149e6e3a387dezanet
valid convolution, l2 regularisation 
Jan 30 2017
d30568b67f87dezanet
add batch normalisation 
Jan 19 2017
1661a4b01cf3dezanet
add rotation again 
Jan 19 2017
17934843cf08dezanet
no augmentation 
Jan 19 2017
3cbe32242a3fdezanet
new data normalization 
Jan 19 2017
4faccc37c63edezanet
add rotation augmentation 
Jan 19 2017
d23e3aa35487Sandro Ivo Sebastiano de Zanet
adaptions for road seg 
Jan 19 2017
46cb8ef0e096dezanet
reunite 
Jan 12 2017
18bb39af0995dezanet
back to net 
Jan 10 2017
95212031368bdezanet
more filters, batchsize10 
Jan 9 2017
fdabe7c466e9dezanet
double filters 
Jan 9 2017
be53ac9f93e7dezanet
dilated test 
Jan 9 2017
b437cb13c929dezanet
last stand 
Dec 22 2016
5c250b837663dezanet
fixed callback 
Dec 5 2016
d867b1af02e4dezanet
flow net needs two outputs 
Dec 5 2016
33bd35298d07dezanet
fixing problems with disparity regression 
Dec 5 2016
6e1c57828bd0Sandro Ivo Sebastiano de Zanet
changed filters 
Nov 30 2016
c192cda5673bdezanet
moar filters 
Nov 29 2016
c7654b962f35dezanet
moved normalise to data.py 
Nov 29 2016
69ae307e4fa1Sandro Ivo Sebastiano de Zanet
introduced dice again 
Nov 29 2016
6fbedcdf73bddezanet
double width 
Nov 29 2016
e8c5878f03a4dezanet
use loss 
Nov 29 2016
a9bb0e588967dezanet
more conv per level 
Nov 29 2016
6b22507eff9fdezanet
reduce k 
Nov 29 2016
e0efb5ec202ddezanet
augmentation failed 
Nov 29 2016
e2d6c8953022dezanet
augmentation 
Nov 29 2016
c82e40047b48dezanet
return correct sized image 
Nov 29 2016
0ba59ba4b315dezanet
moving the data to -100,100 
Nov 29 2016
ad4e6393599fdezanet
no normalisation 
Nov 28 2016
60fe0433734ddezanet
corrected normalisation 
Nov 28 2016
6343f8e1a0a4dezanet
softmax 
Nov 28 2016
314b613f9644dezanet
different normalisation 
Nov 28 2016
cd22810abef6dezanet
reduced batch size 
Nov 28 2016
01fc096a601adezanet
normalisation bug 
Nov 28 2016
e70cf9c4ea39dezanet
try normalisation 
Nov 28 2016
53be0d61f1c1dezanet
no normalisation 
Nov 28 2016
bf61b63df141dezanet
small cleanup 
Nov 28 2016
1da354ec59d6dezanet
remove generator 
Nov 28 2016
8ab6537f0f89dezanet
included batch size 
Nov 28 2016
4bd57a969d57dezanet
callback images generation 
Nov 28 2016
a58311225bc7dezanet
no augmentation 
Nov 28 2016
b594d30c9ec6dezanet
only flip 
Nov 28 2016
f2c75ed5782adezanet
increased number of samples 
Nov 28 2016
4841838b8b00dezanet
reduced amount of augmentation 
Nov 28 2016
3fa3a03b4310dezanet
proper augmentation 
Nov 28 2016
cb46a795235ddezanet
back to unit 
Nov 25 2016
3f644ffa7187dezanet
batchsize=1 
Nov 22 2016
d91c56aad092dezanet
res_u_net with correct bn 
Nov 18 2016
8ffd98d24061dezanet
unit binary cross 
Nov 16 2016
b8f87ac022a0dezanet
res_u_net with binary cross entropy 
Nov 16 2016
c74d007ac3e9dezanet
res_u_net does not perform well 
Nov 16 2016
66191acb18e9dezanet
res_u_net 
Nov 16 2016
27fa74641d07dezanet
removed batchnormalization, it does not work 
Nov 16 2016
c046199225e2dezanet
fixed batch normalisation 
Nov 16 2016
a28cc3dcb3acdezanet
binary crossen-entropy for unit 
Nov 16 2016
87e7b3142e4ddezanet
fixed image output 
Nov 1 2016
2f322b27dc29dezanet
unit returns more than one value 
Nov 1 2016
799ff068b68ddezanet
callback for expected output 
Sep 26 2016
4620bb019d30dezanet
correct callback 
Sep 26 2016
b5ce2c41ac4fdezanet
added disparity learning 
Sep 26 2016
866b8c3e7cb0dezanet
more bug fixing 
Sep 21 2016
d2da32ad15a7dezanet
bug fixing 
Sep 21 2016
fb9f4b57bc2cdezanet
conform image to model 
Sep 21 2016
5699e42c7002dezanet
parentheses in the wrong place 
Sep 21 2016
0c853f128b4bdezanet
callbacks has to be a list 
Sep 21 2016
cf6f2579b172dezanet
fixed epoch_image 
Sep 21 2016
f0a3e1952f53dezanet
Added CurrentSegmentation output to see segmentation per epoch and for testing… 
Sep 20 2016
05e56b601f96dezanet
pretrained weights and batch normalisation 
Sep 13 2016
cb42c810baf5dezanet
better id 
Sep 12 2016
f3186f7635a1dezanet
main function 
Sep 12 2016
1bdca8a95546dezanet
made terminal-accesible 
Sep 12 2016
09f83423a5ccSandro Ivo Sebastiano de Zanet
adaptions to server 
Sep 9 2016
66b5541281dcdezanet
initial commit 
Sep 7 2016
c4science · Help