 nagadomi
|
d7ab10581c
Use clamp
|
%!s(int64=8) %!d(string=hai) anos |
|
nagadomi
|
f65132dadb
Use conv2d instead of nn.SpatialConvolutionMM
|
%!s(int64=8) %!d(string=hai) anos |
|
nagadomi
|
5a3d012f4e
Add random blur method for data augmentation
|
%!s(int64=8) %!d(string=hai) anos |
|
nagadomi
|
c2e4bb4380
Support for multi-thread in training
|
%!s(int64=8) %!d(string=hai) anos |
|
nagadomi
|
d0630d3a20
individual filters and box-only support
|
%!s(int64=9) %!d(string=hai) anos |
|
nagadomi
|
eea4c31d7b
Change parameters
|
%!s(int64=9) %!d(string=hai) anos |
|
nagadomi
|
aaac6ed6e5
Refactor training loop
|
%!s(int64=9) %!d(string=hai) anos |
|
nagadomi
|
c72ec3112b
Add -random_unsharp_mask_rate option for photo
|
%!s(int64=9) %!d(string=hai) anos |
|
nagadomi
|
eaeaef10b7
more flip
|
%!s(int64=9) %!d(string=hai) anos |
|
nagadomi
|
3ea16b3b86
tunable parameters
|
%!s(int64=9) %!d(string=hai) anos |
|
nagadomi
|
4c691b4640
refactor
|
%!s(int64=9) %!d(string=hai) anos |
