Commit Graph

1791 Commits

Author SHA1 Message Date
AlexeyAB 67c91e6bef Minor fix 2020-01-10 14:44:48 +03:00
AlexeyAB 619e39fc71 Fixed bug in [shortcut] 2020-01-10 14:12:39 +03:00
cyy 300406e26a fix macro 2020-01-10 17:47:10 +08:00
cyy 8bd01dd8da
Merge branch 'master' into master 2020-01-10 17:45:35 +08:00
AlexeyAB 2116cba1ed [shortcut] uses activation=linear by default 2020-01-10 00:03:03 +03:00
AlexeyAB f1643cbf2e Added: weights_normalizion = relu or softmax for [shortcut] layer 2020-01-09 23:53:13 +03:00
AlexeyAB 0dbd0e1ef0 compile fix 2020-01-09 20:06:51 +03:00
AlexeyAB 283333b52f Added network resizing for SAM layer 2020-01-09 19:44:08 +03:00
AlexeyAB a66bc8d6ff Resizing Detector: random=1 (/1.4 - x1.4), random=3 (/3.0 - x3.0), random=0 (no resizing). Added param: resize_step=32 to the [net] section: use resize_step=128 for 7 subsampling layers. 2020-01-09 19:25:02 +03:00
AlexeyAB 6878ecc2e2 Minor fix 2020-01-09 13:48:52 +03:00
AlexeyAB cfac917b75 Minor performance improvement 2020-01-09 01:30:57 +03:00
cyy 9d2e6f9a23 fix compiling errors 2020-01-08 10:56:10 +08:00
AlexeyAB 40006b6e9b Compile fix 2020-01-08 02:26:55 +03:00
AlexeyAB 9bd88d7fd7 weighted [shortcut] layer 2020-01-08 02:20:42 +03:00
cyy 2f6bc96156
Merge branch 'master' into master 2020-01-07 16:21:13 +08:00
AlexeyAB e62506629e Another one compile fix 2020-01-07 01:17:28 +03:00
AlexeyAB 25a3d79a7a Compile fix 2020-01-07 00:04:57 +03:00
AlexeyAB c89482c39b Fixed new [shortcut] layer 2020-01-07 00:01:26 +03:00
AlexeyAB c2ce28bbcc Compile fix 2020-01-06 23:38:21 +03:00
AlexeyAB b267e34487 multilayer [shortcut] in progress 2020-01-06 23:27:41 +03:00
AlexeyAB 8bd7dcbd58 Minor fix 2020-01-05 14:04:11 +03:00
AlexeyAB 2652263727 Added flags -benchmark and -benchmark_layers for: ./darknet detector demo 2020-01-05 00:53:03 +03:00
AlexeyAB 3c134dc033 Added fast random functions 2020-01-04 15:47:15 +03:00
AlexeyAB 75a327856b Merge branch 'master' of github.com:AlexeyAB/darknet 2020-01-04 15:44:18 +03:00
AlexeyAB b1c7331483 Added fast random functions 2020-01-04 15:43:55 +03:00
Alexey cd1c7c32af
Update Readme.md 2020-01-03 04:33:28 +03:00
AlexeyAB 0be40dbc92 Update README.md 2020-01-03 00:21:03 +03:00
AlexeyAB 3ae534e228 csresnext50-panet-spp-original-optimal.cfg 2020-01-02 23:53:06 +03:00
AlexeyAB 489ea7fdce Minor fix 2020-01-01 21:38:06 +03:00
AlexeyAB dcfeea30f1 Compile fix 2019-12-29 15:12:22 +03:00
AlexeyAB 7e7964388b Added output CUDA, cuDNN, OpenCV versions 2019-12-29 14:55:12 +03:00
AlexeyAB 2c08a20b22 Added flag -benchmark_layers for: classifier demo, detector test, detector train 2019-12-28 23:03:30 +03:00
AlexeyAB 987448cad8 Minor fix 2019-12-28 20:26:58 +03:00
AlexeyAB 00f0b05047 Added flag -benchmark for: ./darknet classifier demo ... -benchmark 2019-12-28 20:18:15 +03:00
AlexeyAB 114a7f942b Added command line param -dontuse_opencv for training Classifier. Also use GaussianBlur instead of bilateralFilter for blur=1 for training Classifier and Detector. 2019-12-28 18:46:10 +03:00
AlexeyAB b8605bda1e Fixed mosaic=1 cutmix=1 memory leak for Training Classifier 2019-12-27 12:54:20 +03:00
AlexeyAB 35a3870979 counters_per_class affects the objectness too 2019-12-25 20:34:24 +03:00
AlexeyAB b8fe630119 counters_per_class affects the bbox too 2019-12-25 20:30:56 +03:00
AlexeyAB f64746107d Fixed add_bias vs batch_norm on CPU 2019-12-25 00:17:01 +03:00
AlexeyAB 81c5bcea92 Avg-loss will be saved to the 2 images: chart.png & chart_your_cfg_name.png 2019-12-24 22:03:59 +03:00
AlexeyAB 719180e7f1 Fixed swish/mish on CPU 2019-12-24 04:30:22 +03:00
AlexeyAB f81033266e Minor fix 2019-12-23 20:08:16 +03:00
AlexeyAB 0a841f6a22 Fixed strech_sway=1 param, local_avgpool for large size 2019-12-23 02:04:23 +03:00
AlexeyAB af54d95184 Added stretch_sway 2019-12-22 16:00:20 +03:00
AlexeyAB b2749c9e97 Update Readme.md 2019-12-21 14:05:16 +03:00
AlexeyAB 5f6dced3ce Another fix for [batchnorm] layer 2019-12-20 22:49:09 +03:00
AlexeyAB d6eaa2e95f Fixed [batchnorm] layer 2019-12-20 21:37:28 +03:00
AlexeyAB 005714a10c Added [local_avgpool] layer 2019-12-20 20:28:35 +03:00
AlexeyAB e66d3b1bdb Disabled CUDNN_TENSOR_OP_MATH_ALLOW_CONVERSION, and only if groups==1 use CUDNN_TENSOR_OP_MATH 2019-12-20 16:48:31 +03:00
AlexeyAB dc7277f152 Increased speed on Tensor Cores for csresnext50-panet-spp.cfg model 2019-12-20 16:07:34 +03:00