github/DeepFaceLab
1
0
Fork 0
mirror of https://github.com/iperov/DeepFaceLab.git synced 2025-08-22 06:23:20 -07:00
Code Issues Projects Releases Packages Wiki Activity Actions

Commit graph

  • 9c6ca24642 fix update preview samples after disable pretrain Colombo 2020-01-28 13:32:01 +04:00
  • 7386a9d6fd optimized face sample generator, CPU load is significantly reduced Colombo 2020-01-28 12:24:45 +04:00
  • 806dcdc0d3 fix #581 Ork Blutt 2020-01-27 20:35:46 +01:00
  • 80f285067a upd readme Colombo 2020-01-27 20:30:07 +04:00
  • 9b6bb4cd34 fix io not imported (#580) TooMuchFun 2020-01-26 18:56:04 -08:00
  • d6b1409042 fix io not imported TooMuchFun 2020-01-26 05:50:37 -08:00
  • 17337ea8f8 fixed model renaming/deleting Colombo 2020-01-26 14:49:37 +04:00
  • c485e1718a fixes Colombo 2020-01-26 12:56:21 +04:00
  • 76ca79216e Upgraded to TF version 1.13.2 Colombo 2020-01-25 21:58:19 +04:00
  • a3dfcb91b9 1 Colombo 2020-01-25 18:18:06 +04:00
  • 6a3b0feb22
         Update Model.py zakaria47fs 2020-01-24 13:26:05 +01:00
  • af20983d31
         Update Model.py zakaria47fs 2020-01-24 13:15:38 +01:00
  • f7f4e44a98 1 Colombo 2020-01-24 11:51:54 +04:00
  • be5df1e269 fix Colombo 2020-01-23 22:46:45 +04:00
  • 1f0c91f053 add SampleGeneratorFaceTemporal.py Colombo 2020-01-23 19:09:23 +04:00
  • afcfa9c69e fix io.input_skip_pending for colab Colombo 2020-01-23 10:20:29 +04:00
  • 591bba2c1e fix clipgrad option Colombo 2020-01-23 09:30:43 +04:00
  • 967fbdd814 BREAKING CHANGES !!! Colombo 2020-01-22 22:23:55 +04:00
  • f1a3150ce9 1 Colombo 2020-01-22 22:14:31 +04:00
  • 869c5e8dd0 1 Colombo 2020-01-22 18:58:02 +04:00
  • af3306a85a 1 Colombo 2020-01-22 18:56:00 +04:00
  • 0ce96875a1 1 Colombo 2020-01-22 18:49:32 +04:00
  • 2d61b715de 1 Colombo 2020-01-22 18:27:19 +04:00
  • afafd89982 1 Colombo 2020-01-22 17:23:08 +04:00
  • 3924ef2bf5 1 Colombo 2020-01-22 17:21:33 +04:00
  • 60804ca3ba 1 Colombo 2020-01-22 13:41:05 +04:00
  • 0c93b89e87 1 Colombo 2020-01-22 10:46:35 +04:00
  • beed145d29 1 Colombo 2020-01-22 10:29:17 +04:00
  • 9797a70fd3 1 Colombo 2020-01-21 21:05:29 +04:00
  • 56de1d9fa5 1 Colombo 2020-01-21 18:50:52 +04:00
  • 38b85108b3 DFL-2.0 initial branch commit Colombo 2020-01-21 18:43:39 +04:00
  • 52a67a61b3 Sample loader : back to serial one core loader DFL-1.0 Colombo 2020-01-11 23:20:39 +04:00
  • bbe81b20af - Colombo 2020-01-08 15:12:43 +04:00
  • b8182ae42b remove lr_dropout for plaidml backend Colombo 2020-01-08 11:11:33 +04:00
  • d3e6b435aa fixes and optimizations Colombo 2020-01-07 13:45:54 +04:00
  • 842a48964f fix Colombo 2020-01-05 14:21:34 +04:00
  • ea33541177 fix Colombo 2020-01-05 13:58:25 +04:00
  • 21b25038ac optimized sample generator Colombo 2020-01-05 11:53:31 +04:00
  • b5c234dac3 no_preview for model Colombo 2020-01-04 23:52:12 +04:00
  • 2429d28737 optimize memory usage Colombo 2020-01-04 23:51:33 +04:00
  • f95682f616 fix Colombo 2020-01-02 21:19:19 +04:00
  • 5efb430f2a +WScaleConv2DLayer Colombo 2020-01-01 11:15:55 +04:00
  • efd8c16daa avatar: avatar_type is now only head by default Colombo 2019-12-29 19:37:59 +04:00
  • 47e539ccdd fix extract unaligned faces Colombo 2019-12-29 19:36:34 +04:00
  • 818a2d65f0 fix long load when using various gpus in the same DFL folder Colombo 2019-12-29 18:47:55 +04:00
  • 540650c610 fix faceset enhancer for frames that contain edited mask Colombo 2019-12-29 18:10:05 +04:00
  • 844cfa48b0 还原 SAEHD 模型参数 cszdlt 2019-12-29 01:18:27 +08:00
  • e612c1ab06 降低学习率 cszdlt 2019-12-29 00:59:23 +08:00
  • 36bff495b8 Merge branch 'master' of github.com:cszdlt/DeepFaceLab cszdlt 2019-12-29 00:54:42 +08:00
  • c104ba6239 还原SAEHD模型参数 cszdlt 2019-12-29 00:53:53 +08:00
  • f62fff7f0f
         Merge branch 'master' into master cszdlt 2019-12-29 00:48:34 +08:00
  • 28549dc153 SAEHD:optimized architecture, you have to restart training Colombo 2019-12-28 16:50:33 +04:00
  • 4e67278c66 fix 0 faces detected in manual extractor Colombo 2019-12-28 16:45:40 +04:00
  • 3849172a43 add GeLU Colombo 2019-12-27 21:25:59 +04:00
  • 0833a38bb9 Quick96: optimized architecture Colombo 2019-12-27 21:06:45 +04:00
  • a347bb4f15 change doc about builds Colombo 2019-12-27 14:29:27 +04:00
  • 7804e0bd7c add try except Colombo 2019-12-27 13:36:01 +04:00
  • df8cf35c75 FacesetEnhancer now asks to merge aligned_enhanced/ to aligned/ Colombo 2019-12-27 11:09:15 +04:00
  • d46fb5cfd3 fixed mask editor Colombo 2019-12-26 21:27:10 +04:00
  • 3be223a265 fix MaskEditor Colombo 2019-12-25 11:13:52 +04:00
  • ac1913c070 fix Colombo 2019-12-24 14:07:06 +04:00
  • 3feb386a8e fix for colab Colombo 2019-12-24 11:20:32 +04:00
  • fe724c8322
         Update Model.py cszdlt 2019-12-24 10:25:54 +08:00
  • c482b2cd07 upd reaedme Colombo 2019-12-23 19:16:33 +04:00
  • b93ae56a08 upd readme Colombo 2019-12-23 19:15:59 +04:00
  • 8c17dd85f4 upd readme Colombo 2019-12-23 19:06:30 +04:00
  • 94c99b429d fix Colombo 2019-12-23 15:17:13 +04:00
  • c1612c5553 fixes Colombo 2019-12-23 14:57:47 +04:00
  • 7174dc835a fixes Colombo 2019-12-22 21:44:28 +04:00
  • 7e609542db fixes Colombo 2019-12-22 21:17:34 +04:00
  • 9ed0111824 fix Colombo 2019-12-22 19:10:57 +04:00
  • 754d6c385c refactorings Colombo 2019-12-22 19:00:59 +04:00
  • e0e8970ab9 fix PackedFaceset Colombo 2019-12-22 15:58:46 +04:00
  • dd45b7dacc cleaning Colombo 2019-12-22 15:02:53 +04:00
  • 8153e90ba3 refactoring Colombo 2019-12-22 14:05:41 +04:00
  • 246129a964 change samples.pak to faceset.pak Colombo 2019-12-22 13:32:19 +04:00
  • 9312a3c334 Extractor: 2nd pass now faster on frames where faces are not found Colombo 2019-12-22 10:46:44 +04:00
  • 8305832a27 fix Colombo 2019-12-21 23:27:51 +04:00
  • 50f892d57d all models: removed options 'src_scale_mod', and 'sort samples by yaw as target' If you want, you can manually remove unnecessary angles from src faceset after sort by yaw. Colombo 2019-12-21 23:16:55 +04:00
  • 8866dce22e added 4.2.other) data_src util faceset metadata save.bat saves metadata of data_src\aligned\ faces into data_src\aligned\meta.dat Colombo 2019-12-20 15:03:17 +04:00
  • 5e6a1f5249 fix Colombo 2019-12-20 14:16:44 +04:00
  • 021bb6d128 SAEHD: lr_dropout now as an option, and disabled by default Colombo 2019-12-20 12:04:44 +04:00
  • 951942821d 1 Colombo 2019-12-20 11:04:26 +04:00
  • cf27558e88 SAEHD: lr_dropout now as an option, and disabled by default Colombo 2019-12-20 10:39:05 +04:00
  • 64021b9c62 more stable and precise version of face transformation matrix. fixed bleeding mask on some samples Colombo 2019-12-20 10:30:49 +04:00
  • 068c7d0d55 temporary revert last fixes Colombo 2019-12-20 10:21:59 +04:00
  • dd1d5e8909 improved face align, More stable and precise version of the face transformation matrix. Now full_faces are aligned with the upper and lateral boundaries of the frame, result: fix of cutted mouth, increase area of the cheeks of side faces before/after https://i.imgur.com/t9IyGZv.jpg therefore, additional training is required for existing models. Optionally, you can re-extract dst faces of your project, if they have problems with cutted mouth or cheeks. Colombo 2019-12-19 18:33:04 +04:00
  • 9e9dc364c9 temporary revert fix Colombo 2019-12-19 15:46:50 +04:00
  • 853a056769 more stable and precise version of face transformation matrix Colombo 2019-12-19 15:25:06 +04:00
  • c04740eac1 1 Colombo 2019-12-17 09:53:17 +04:00
  • 28b948888c fix extractor for some systems Colombo 2019-12-17 09:52:48 +04:00
  • 719f5fb790
         Bump tensorflow from 1.12.0 to 1.15.0 dependabot[bot] 2019-12-16 21:54:22 +00:00
  • 5e4faa6948
         Bump tensorflow-gpu from 1.12.0 to 1.15.0 dependabot[bot] 2019-12-16 20:41:40 +00:00
  • 71ebf06c89 SAEHD,Quick96: improved model generalization, overall accuracy and sharpness by using new 'Learning rate dropout' technique from paper https://arxiv.org/abs/1912.00144 An example of a loss histogram where this function is enabled after the red arrow: https://i.imgur.com/3olskOd.jpg Colombo 2019-12-15 15:53:06 +04:00
  • c866448645 clean Colombo 2019-12-11 22:34:22 +04:00
  • d4745b5cf8 added sort by absdiff This is sort method by absolute per pixel difference between all faces. options: Sort by similar? ( y/n ?:help skip:y ) : if you choose 'n', then most dissimilar faces will be placed first. Colombo 2019-12-11 22:33:49 +04:00
  • e8673e3fcc nothing interesting Colombo 2019-12-11 22:33:23 +04:00
  • 154820a954 ConvertMasked: fix color bug for superresolution Colombo 2019-12-06 16:49:27 +04:00
  • 30ef74cad3
         Update Model.py cszdlt 2019-12-03 15:21:55 +08:00
  • b62eb0da38 modify learn rate to 1e-5 cszdlt 2019-12-02 23:39:03 +08:00