Commit Graph

757 Commits

Author SHA1 Message Date
381f90ebb8 Fixed CUDA error checking 2019-01-29 13:46:30 +03:00
2790464de1 Another compile fix 2019-01-29 00:11:32 +03:00
ae8a8e6016 Compile fix 2019-01-29 00:05:08 +03:00
640bdbc063 LSTM, RNN, GRU - use connected_layer that uses cuDNN. Fixed CRNN for conv-layer with cuDNN. 2019-01-28 23:50:51 +03:00
0e1f3eaf35 Fixed DLL/SO 2019-01-28 20:32:30 +03:00
3692c174c5 Compile fix 2019-01-28 20:25:14 +03:00
110b5240a4 Fixed LSTM-layer 2019-01-28 20:22:14 +03:00
85b99872cb Use non-default stream for all CUDA-functions 2019-01-28 20:19:26 +03:00
00b87281f3 Fixed RNN (RNN, GRU, LSTM) with cuDNN (batch-norm) 2019-01-27 03:42:44 +03:00
9576cd4d89 Fixed memory allocation 2019-01-26 23:25:09 +03:00
090d934c0f Minor speedup on CPU 2019-01-26 19:12:46 +03:00
630f441e08 Minor CPU speedup - i7 6500K: 1000ms (AVX=1) instead of 1500ms (old AVX=1) and 2000ms (AVX=0) 2019-01-26 02:54:41 +03:00
da044776d1 Merge pull request #2282 from davidssmith/master
add LSTM layer
2019-01-24 20:19:57 +03:00
a7366a5a0a Compile fix for CC < 7.3 2019-01-24 20:19:01 +03:00
5e778cd91e add LSTM layer 2019-01-23 22:02:09 -06:00
2d3220cef5 Look at wmma::bmma_sync(), bmmaBitOpXOR, bmmaAccumulateOpPOPC 2019-01-23 00:35:44 +03:00
8960fbfb3f gcc on OSX required explicit return value for empty (char *) in detection_to_json 2019-01-22 10:17:42 -08:00
2cd37ec73e Another minor fix 2019-01-22 17:31:07 +03:00
46be08db37 Minor fix 2019-01-22 16:23:44 +03:00
81f7fc2c7b Fixed network resize memory allocation 2019-01-21 15:32:57 +03:00
226322523e Fixed calc_anchors 2019-01-21 14:54:51 +03:00
17019854c3 XNOR minor fix 2019-01-19 03:18:50 +03:00
6e99e852ff Network resize is fixed 2019-01-18 23:30:10 +03:00
0e022d0912 Fixed timer 2019-01-18 21:29:06 +03:00
3a51f4af74 Experimental repack 2019-01-18 19:52:11 +03:00
bf6b40f4e9 Another CUDA performance improvements 2019-01-18 16:29:54 +03:00
5343aa4235 CUDA minor performance improvement 2019-01-16 18:08:11 +03:00
4c05166215 Temporary experimental XNOR on GPU (repack channels) 2019-01-16 02:43:44 +03:00
83620d8b28 Training minor fix 2019-01-14 14:42:48 +03:00
d2b5e142b3 Minor fix 2019-01-14 13:56:42 +03:00
a590609b1b Compile fix 2019-01-12 13:50:51 +03:00
70260cda24 Added -mjpeg_port 8090 for Training. Loss-chart saved to chart.png even when used -dont_show.
mAP is calculated during training even without OpenCV.
2019-01-12 13:45:37 +03:00
5c55137687 Minor fix 2019-01-12 02:56:30 +03:00
eb64e2b13c Added SO_REUSEADDR and SO_REUSEPORT for JSON & MJPEG streams 2019-01-11 23:11:25 +03:00
937306f0c9 Fixed JSON-stream and MJPEG-stream on Windows and Linux. 2019-01-11 22:39:29 +03:00
95904cf65a Bad memory allocation
the variable [network->seen is declared uint64_t* ](https://github.com/AlexeyAB/darknet/blob/master/include/darknet.h#L519), but constructing function allocates only 4 bytes for (int) here
2019-01-11 14:56:03 +02:00
041cffd9b8 Compile fix 2019-01-11 01:05:44 +03:00
580b2fb83c Fixed ./darknet detect ... 2019-01-11 00:52:19 +03:00
97038fefa6 Fixed bug with saving training Loss-Chart 2019-01-09 21:47:37 +03:00
5ee166032a Use detector test...-out result.json -dont_show < train.txt to save results to JSON-file 2019-01-09 17:49:37 +03:00
596311e2c7 Another compile fix ) 2019-01-09 13:53:33 +03:00
63a2f005aa Compile fix 2019-01-09 13:46:53 +03:00
e6eba22adb Graceful shutdown closes sockets 2019-01-09 03:20:29 +03:00
c31a245665 Experimental. Use -json_port 8070 to allow connection to the port:8070 to get json-stream 2019-01-09 00:11:59 +03:00
ead84c8ba2 Now you can un-comment several lines for saving images of cropped objects to /result_img/ directory 2019-01-08 18:34:59 +03:00
3ff5084590 Added include/darknet.h 2019-01-06 23:51:38 +03:00
c56931dd75 Compile fix 2019-01-06 16:00:00 +03:00
c75fbb5f2e Minor fix 2019-01-06 15:45:10 +03:00
48d461f9bd Temporary experimental XNOR improvements on CPU 2019-01-04 23:19:45 +03:00
527578744b Minor fix 2018-12-29 22:56:38 +03:00