e58197ca22
initial code to extract umdfaces.io dataset and train pose estimator
2019-04-23 08:14:09 +04:00
046649e6be
update == 04.20.2019 == ( #242 )
...
* superb improved fanseg
* _
* _
* added FANseg extractor for src and dst faces to use it in training
* -
* _
* _
* update to 'partial' func
* _
* trained FANSeg_256_full_face.h5,
new experimental models: AVATAR, RecycleGAN
* _
* _
* _
* fix for TCC mode cards(tesla), was conflict with plaidML initialization.
* _
* update manuals
* _
2019-04-20 08:23:08 +04:00
2276f5eb73
fix zoom keys
2019-04-08 17:35:06 +04:00
8561c1a931
fix MaskEditor, upd manual ru
2019-04-08 14:13:36 +04:00
7c930121e9
MaskEditor: added window zoom
2019-04-08 10:30:59 +04:00
a88ee7d093
fix
2019-04-06 17:48:33 +04:00
43d68e1307
MaskEditor: added prev/next frames preview
2019-04-06 16:22:48 +04:00
9f68b25987
Extractor: added warning messages if aligned folder contains files that will be deleted.
2019-04-06 10:01:59 +04:00
cfae9f9fd8
trainer: fix for ctrl-c in --no-preview mode
2019-04-05 21:50:46 +04:00
fe40a46933
limiting converter subprocesses to max 6
2019-04-05 21:43:24 +04:00
406ef91851
fix mask editor
2019-04-05 14:42:11 +04:00
d3ac88b799
fix MaskEditorTool
2019-04-05 09:32:04 +04:00
de8d75b4f7
fix mask editor,
...
upd manuals
2019-04-04 22:19:56 +04:00
0fa9d269ec
fixes
2019-04-04 16:34:45 +04:00
5ac7e5d7f1
changed help message for pixel loss:
...
Pixel loss may help to enhance fine details and stabilize face color. Use it only if quality does not improve over time.
SAE:
previous SAE model will not work with this update.
Greatly decreased chance of model collapse.
Increased model accuracy.
Residual blocks now default and this option has been removed.
Improved 'learn mask'.
Added masked preview (switch by space key)
Converter:
fixed rct/lct in seamless mode
added mask mode (6) learned*FAN-prd*FAN-dst
added mask editor, its created for refining dataset for FANSeg model, and not for production, but you can spend your time and test it in regular fakes with face obstructions
2019-04-04 10:22:53 +04:00
85c01e3b4a
Converter: added Apply super resolution? (y/n skip:n) : , Enhance details by applying DCSCN network.
...
refactorings
2019-03-28 21:50:27 +04:00
689aefeb2f
Converter: fix for input images with ch > 3
2019-03-28 11:12:19 +04:00
fedeee4395
Colab: added ability to run programs after N secs in trainer
2019-03-26 19:06:19 +04:00
009af42617
fix converter issue #205
2019-03-26 18:23:39 +04:00
96de328221
upd for Colab ( #206 )
...
* changes for google colab
* 1
* 1
* 1
* 1
2019-03-26 16:09:22 +04:00
b03b147bae
refactoring
2019-03-26 11:09:44 +04:00
131b2b5c79
after save loss string now shows averaged value since last save
2019-03-25 18:14:04 +04:00
37505d88e3
old SAE model will not work with this update.
...
Fixed bug when SAE can be collapsed during a time.
SAE: removed CA weights and encoder/decoder dims.
added new options:
Encoder dims per channel (21-85 ?:help skip:%d)
More encoder dims help to recognize more facial features, but require more VRAM. You can fine-tune model size to fit your GPU.
Decoder dims per channel (11-85 ?:help skip:%d)
More decoder dims help to get better details, but require more VRAM. You can fine-tune model size to fit your GPU.
Add residual blocks to decoder? (y/n, ?:help skip:n) :
These blocks help to get better details, but require more computing time.
Remove gray border? (y/n, ?:help skip:n) :
Removes gray border of predicted face, but requires more computing resources.
2019-03-24 15:35:02 +04:00
673d4915d9
VideoEd: added comment
2019-03-23 18:28:50 +04:00
668a3e23a1
frame extraction: fixed pixel format for PNG. Added PNG/JPG option.
2019-03-23 18:27:38 +04:00
7a0cc56603
extractor: fixes, optimizations,
...
manual extractor: added 'a' option to switch accuracy mode
2019-03-21 18:56:32 +04:00
d1676dea06
refactorings, optimizations
2019-03-21 11:58:54 +04:00
a3df04999c
removing trailing spaces
2019-03-19 23:53:27 +04:00
1f569117c8
accurate landmark extractor now by default
2019-03-19 14:01:15 +04:00
a0c1ce3e7e
final sort: increased speed
2019-03-19 13:45:26 +04:00
7c1ea1a617
removing dflimg image_to_face_mat
2019-03-18 14:42:10 +04:00
00c6a0cdaf
accurate landmarks extraction now as an option
2019-03-18 11:24:41 +04:00
bf831331e6
fix
2019-03-18 10:58:03 +04:00
b8efb4cbba
enhancing landmarks extractor by using s3fd second pass inside second pass,
...
it will be x2 slower, but time will be saved due to more images will be marked properly
works on 2GB+
2019-03-18 10:25:24 +04:00
1d56585f33
fix ability for auto re-extract dst faces, after this update DFL torrent extractor scripts will not work.
2019-03-18 01:12:04 +04:00
b03d62a29b
extractor: added ability to re-extract dst faces with transferring all necessary meta data
2019-03-18 00:43:07 +04:00
4ff88865ce
refactoring
2019-03-16 21:29:56 +04:00
f3b343c0e5
small fixed and refactorings
2019-03-16 20:55:51 +04:00
7c66503382
sorter: fix final sort
2019-03-16 09:20:55 +04:00
ea5fe28d08
fix final sorter
2019-03-15 23:41:34 +04:00
3549272244
fix for final
2019-03-15 11:03:37 +04:00
9823421a44
added transparent mask to draw_landmarks
2019-03-14 12:16:21 +04:00
5bd3f875fa
fix
2019-03-14 10:50:53 +04:00
6e59f1a150
Sorter: new better blur estimator
2019-03-14 10:38:56 +04:00
62af1b6f16
S3FD : filtering faces < 40pix.
2019-03-14 01:23:39 +04:00
97b6fabaab
change 'epoch' to 'iter',
...
added timestamp prefix to training string
2019-03-12 19:23:52 +04:00
893354b870
fix S3FD for plaidML, fix S3FD warnings on Lambda layers.
2019-03-11 09:32:00 +04:00
fbf39d2727
added new extractor: S3FD,
...
all extractors now produce less false-positive faces
2019-03-10 23:18:10 +04:00
168d48b119
extractor: increased speed of 1st pass by DLIB for >= 4GB vram cards,
...
added 'h' key for hide the help in manual extractor
2019-03-08 17:01:11 +04:00
6094e5ead8
added Util.recover_original_aligned_filename
2019-03-04 10:57:44 +04:00
iperov