 nagadomi
|
0fe21eef70
Update random erasing; Support x padding
|
7 years ago |
|
nagadomi
|
93cd40a53c
merge random erasing
|
7 years ago |
|
nagadomi
|
ed12b43477
Change downsampling filters in pairwise_scale()
|
8 years ago |
|
nagadomi
|
c6e3a68974
Optimize flip in user method when -patchs is small
|
8 years ago |
|
nagadomi
|
6220afc31a
Fix a bug in random negate
|
9 years ago |
|
nagadomi
|
b066761cdc
Add data augmentation for user method
|
9 years ago |
|
nagadomi
|
a72af8cfef
Use 'image' module because the deadlock problem has fixed
|
9 years ago |
|
nagadomi
|
d7ab10581c
Use clamp
|
9 years ago |
|
nagadomi
|
f65132dadb
Use conv2d instead of nn.SpatialConvolutionMM
|
9 years ago |
|
nagadomi
|
5a3d012f4e
Add random blur method for data augmentation
|
9 years ago |
|
nagadomi
|
c2e4bb4380
Support for multi-thread in training
|
9 years ago |
|
nagadomi
|
d0630d3a20
individual filters and box-only support
|
9 years ago |
|
nagadomi
|
eea4c31d7b
Change parameters
|
10 years ago |
|
nagadomi
|
aaac6ed6e5
Refactor training loop
|
10 years ago |
|
nagadomi
|
c72ec3112b
Add -random_unsharp_mask_rate option for photo
|
10 years ago |
|
nagadomi
|
eaeaef10b7
more flip
|
10 years ago |
|
nagadomi
|
3ea16b3b86
tunable parameters
|
10 years ago |
|
nagadomi
|
4c691b4640
refactor
|
10 years ago |
