Commit Graph

53 Commits

Author SHA1 Message Date
8b5344ee2d Added BFLOPs output for network configurations 2018-05-14 13:34:40 +03:00
9bae70b225 Accelerated by another 5% using FP16/32 Batch-norm for Tensor Cores. 2018-04-17 02:51:11 +03:00
eb9c88ef73 Fixed bug in Tensor Cores V100 (1. Desc in Batch norm, 2. Manually selected algo).
Also fixed time measure on Linux for multi-threading.
2018-04-15 01:51:21 +03:00
9fbd1936eb small_object can be used in [net] in cfg-file for objects 1x1 pix on FullHD image 2018-03-30 14:32:41 +03:00
b659d93a4a Changed cmd and sh files. Added new cfg files for old Yolo v2. 2018-03-28 23:23:37 +03:00
d9ae3dd681 Added Yolo v3 2018-03-28 02:59:03 +03:00
537d135feb Improve training performance - batch-norm using cuDNN. 2018-03-20 02:16:51 +03:00
5e448c00f3 Added Focal loss. Use focal_loss=1 in the [region] layer in your cfg-file.
alpha = 0.5, gamma = 2
2018-03-05 19:59:29 +03:00
fe44d3d0f2 Fixed reorg_layer. Old reorg can be used as [reorg_old] in the cfg-file. 2018-03-03 18:09:46 +03:00
cad4d1618f Added support for Tensor Cores CC >= 7.0 (V100). For FP16/32 (mixed precision) define CUDNN_HALF should be used. 2018-02-25 16:29:44 +03:00
dda993f3dd Use half_float16 instead of float32 if defined both CUDNN and CUDNN_HALF. Use Tensor Cores. 2018-02-22 22:54:40 +03:00
9ff40baa4f Fixes for training Yolo on small objects 2018-02-08 01:04:40 +03:00
9ac78d8b84 Fine tuning, use stopbackward=1 in the cfg-file in that layer where Backward should be stopped. 2018-01-05 00:58:52 +03:00
0d6b107ed2 hey 2016-11-15 22:53:58 -08:00
4b60afcc64 stuff and things 2016-11-07 23:42:19 -08:00
252e3b1916 :charizard: :charizard: :charizard: 2016-11-05 17:27:31 -07:00
c7a700dc22 new font strategy 2016-11-05 14:09:21 -07:00
352ae7e65b ADAM 2016-10-26 08:35:44 -07:00
d8adaf8ea6 tree stuff 2016-10-21 13:16:43 -07:00
481b57a96a So I have this new programming paradigm....... 2016-09-24 23:12:54 -07:00
5c067dc447 good chance I didn't break anything 2016-09-12 13:55:20 -07:00
8f1b4e0962 updates and things 2016-09-01 16:48:41 -07:00
845ab75796 some more stuff 2016-08-05 15:27:07 -07:00
9361292c42 updates 2016-07-19 14:50:01 -07:00
afb8b4f98b CVPR prep 2016-06-22 21:46:32 -07:00
8a767f1066 stuff for carlo 2016-06-06 15:48:52 -07:00
4625a16ffd tactics 2016-06-06 13:22:45 -07:00
ec3d050a76 hope i didn't break anything 2016-06-02 15:25:24 -07:00
13209df7bb art, cudnn 2016-05-13 11:59:43 -07:00
c7b10ceadb so much need to commit 2016-05-06 16:25:16 -07:00
cff59ba135 go updates 2016-03-16 04:30:48 -07:00
d1965bdb96 Go 2016-03-13 23:18:42 -07:00
16d06ec0db stuff 2016-02-29 13:54:12 -08:00
c604f2d994 ... 2016-01-31 15:52:03 -08:00
913d355ec1 lots of stuff 2016-01-28 12:30:38 -08:00
1578ec70d7 idk 2016-01-18 15:40:14 -08:00
db0397cfaa shortcut layers, msr networks 2015-12-14 11:57:10 -08:00
0305fb4d99 Some changes 2015-11-26 11:48:01 -08:00
0cd2379e2c some changes 2015-11-14 12:34:17 -08:00
8c5364f585 New YOLO 2015-11-09 11:31:39 -08:00
8fd18add6e CVPR Experiments 2015-11-03 19:23:42 -08:00
c40cdeb402 lots of comparator stuff 2015-10-09 12:50:43 -07:00
40cc104639 idk 2015-09-28 14:32:28 -07:00
fed6d6e31d idk man 2015-09-21 18:18:51 -07:00
c53e03348c yolo working w/ regions 2015-09-16 14:12:10 -07:00
8bcdee8658 Some bug fixes, random stuff 2015-09-01 11:22:03 -07:00
9d42f49a24 changing data loading 2015-08-24 18:27:42 -07:00
a47874a7e8 nochange option in crop 2015-07-13 23:25:08 -07:00
8561e49b5a add avgpool layer 2015-07-13 15:04:21 -07:00
75db98db25 normalization layer 2015-07-10 16:38:30 -07:00