Commit Graph

138 Commits

Author SHA1 Message Date
dc7f8a32ae mAP calculation during training, if is used flag -map 2018-12-09 18:18:47 +03:00
25f65f6878 Added fast_binarize_weights_gpu() 2018-11-05 22:38:35 +03:00
c0e2512af2 Activation improvement, more robust timer. 2018-09-27 23:10:54 +03:00
7dd97537fb XNOR-net tiny-yolo_xnor.cfg ~2x faster than cuDNN on CUDA (nVidia GPU Maxwell) 2018-09-22 02:01:14 +03:00
03e95320a1 XNOR coalesced memory access, and avoid bank conflicts 2018-09-17 23:39:25 +03:00
a9fef1bd66 Bug fixes. Tested im2col_cpu_custom_transpose - bad way. 2018-08-11 00:26:53 +03:00
d6162af210 Optimized on CPU: gemm_bin, im2col, activation, transpose 2018-08-09 02:31:36 +03:00
a284a7da8d Try to use avx_hs() - slow and requires alignment 4096 bits < (l.size*l.size*l.c)
May be faster only from 8192 bits and more.
2018-08-08 19:08:58 +03:00
0a326e7afe XNOR-net on CPU AVX2 2018-08-08 02:45:47 +03:00
cfc5fedbb6 Just used spaces for indents instead of Tabs 2018-07-10 23:29:15 +03:00
83b2ae66e1 Expose YOLODLL_API layer* get_network_layer(network* net, int i); 2018-06-10 09:45:39 +08:00
1c05ebf522 Minor fix 2018-06-07 03:39:30 +03:00
6c54f5ffb2 Show inconsistent information if it is present in .cfg and .names files 2018-06-07 02:20:44 +03:00
1e512b2308 Letter box is disabled in the darknet.py by defualt 2018-06-04 22:11:05 +03:00
51293290fa Fix error in allocation size 2018-05-31 13:11:19 +03:00
160eddddc4 Minor fix 2018-05-27 16:53:42 +03:00
eef9f8e5bb Minor fixes 2018-05-23 18:35:08 +03:00
89c11f83ed Fixed darknet.py - uses batch=1 by default 2018-05-16 15:05:46 +03:00
89354d0a0c Fixed memory leaks. And fixes for Web-camera and IP-camera. 2018-05-04 23:52:05 +03:00
e18a303d1c Fixed utils.c for short lines. 2018-04-21 15:19:58 +03:00
fc496d52bf Yolo can be used from Python 2.x using /darknet.py on Linux or darknet/x64/darknet.py on Windows 2018-04-12 22:32:57 +03:00
4d9a2bdac6 Minor fixes 2018-04-11 23:04:23 +03:00
943f6e874b Added Focal Loss to yolo-layer 2018-04-05 23:27:02 +03:00
be9d971ddb Compile error fix 2018-04-03 16:56:53 +03:00
1b2c70f82a Detection is accelerated by 7 percent (fused conv and batch_norm layers) 2018-04-03 16:42:00 +03:00
d0039f6dfd Temporary fix 2018-03-29 02:49:40 +03:00
3fc3fd0f1f Fixed training, calculation mAP and anchors for Yolo v3 2018-03-29 02:38:35 +03:00
d9ae3dd681 Added Yolo v3 2018-03-28 02:59:03 +03:00
371f21171a shortcut_layer resize for random=1 2018-03-16 01:49:01 +03:00
880cf187d8 Fixed multi-GPU training for Tensor Cores 2018-03-09 19:44:46 +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
f558d5c39c Fix 2018-02-22 23:16:36 +03:00
033e934ce8 If there is excessive GPU-RAM consumption by CUDNN then then do not use Workspace 2018-02-21 19:14:01 +03:00
e4ab47dfce Optimized resizing of region_layer for random=1 2018-02-21 15:35:09 +03:00
4b0be8c701 Optimized resizing of network for random=1 2018-02-21 15:06:11 +03:00
db3a3c54c5 Some fixes 2018-02-03 15:35:13 +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
2baa7bde54 Fixed classifier for AlexNet and Resnet50 2017-09-15 20:31:48 +03:00
576dbe12e6 Fixed memory release 2017-07-25 14:28:17 +03:00
5bc62b14e0 Memory leak fixed 2017-07-03 14:12:52 +03:00
0d6b107ed2 hey 2016-11-15 22:53:58 -08:00
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
73f7aacf35 better multigpu 2016-09-20 11:34:49 -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
e7072b8489 checkpoint 2016-06-20 13:18:59 -07:00
8322a58cf6 hate warnings 2016-06-14 11:30:28 -07:00