608 Commits (0c07837397e791c750e81857e214d315ea24199f)
 

Author SHA1 Message Date
AlexeyAB 0c07837397 Added resnet152_yolo.cfg with focal_loss=1 and stopbackward=1 in the last shortcut layer. 7 years ago
AlexeyAB 5e448c00f3 Added Focal loss. Use focal_loss=1 in the [region] layer in your cfg-file. 7 years ago
AlexeyAB f0bc634a74 Fixed saving video file 7 years ago
Alexey 09fdb67d66
Update Readme.md 7 years ago
AlexeyAB c123f320fc Show avg-loss chart during Training, if compiled with OpenCV. Use -dont_show to disable. 7 years ago
AlexeyAB 48586c8d4d Compile fix 7 years ago
AlexeyAB 74f98da211 Show IoU, save anchors to file.Show anchors in the window if used -show 7 years ago
AlexeyAB f39160f6e8 Added: calc_anchors 7 years ago
AlexeyAB fe44d3d0f2 Fixed reorg_layer. Old reorg can be used as [reorg_old] in the cfg-file. 7 years ago
AlexeyAB e16c901163 minor fix 7 years ago
AlexeyAB be9ccb7d7f Added F1 score to accuracy statistic 7 years ago
Alexey 1e84a2804e
Update Readme.md 7 years ago
Alexey 671735fb00
Update Readme.md 7 years ago
AlexeyAB 7caf345d36 One more fix 7 years ago
AlexeyAB 18d5670d64 Compile fix 7 years ago
AlexeyAB 5cd147031b Added param -dont_show to detect objects on servers without display, like Amazon EC2 7 years ago
AlexeyAB 35a66b73a2 Changed class to class_id 7 years ago
Alexey 17b22d7ce7
Update Readme.md 7 years ago
AlexeyAB 490d02505b Fixed VideoWriter 7 years ago
Alexey e96a454ca1
Update Readme.md 7 years ago
AlexeyAB 48d713fd49 Fine-tuning densenet201_yolo.cfg and resnet50_yolo.cfg 7 years ago
AlexeyAB cad4d1618f Added support for Tensor Cores CC >= 7.0 (V100). For FP16/32 (mixed precision) define CUDNN_HALF should be used. 7 years ago
AlexeyAB 85eafd3d59 Added partial.cmd 7 years ago
AlexeyAB 3ff4797b4c Added gen_anchors.py 7 years ago
Alexey 175e47317e
Update Readme.md 7 years ago
AlexeyAB cd2bdec090 Updated to CUDA 9.1. And fixed no_gpu dependecies. 7 years ago
AlexeyAB 6332ea99ab one more fix 7 years ago
AlexeyAB b2b5756d86 Added __float2half_rn() and __half2float() 7 years ago
AlexeyAB f558d5c39c Fix 7 years ago
AlexeyAB dda993f3dd Use half_float16 instead of float32 if defined both CUDNN and CUDNN_HALF. Use Tensor Cores. 7 years ago
AlexeyAB 033e934ce8 If there is excessive GPU-RAM consumption by CUDNN then then do not use Workspace 7 years ago
AlexeyAB e4ab47dfce Optimized resizing of region_layer for random=1 7 years ago
AlexeyAB 4b0be8c701 Optimized resizing of network for random=1 7 years ago
AlexeyAB 068090c3a4 Minor fix 7 years ago
AlexeyAB 2972f8de8e Fixed Tracking using SparsePyrLKOpticalFlow on CPU 7 years ago
Alexey 5f74aeb224
Merge pull request #398 from HowardStark/master 7 years ago
Howard Stark d8b59fcafe
Changed definition location to adhere to compatability promises 7 years ago
Howard Stark 892860e84e
Added CPU Optical Flow 7 years ago
Alexey 4a7a076d09
Update Readme.md 7 years ago
AlexeyAB 8474f491f2 Minor fix 7 years ago
Alexey 1106f5325b
Update Readme.md 7 years ago
AlexeyAB 82f630cac7 Added param -http_port 8090 to show MJPEG-stream in the WebBrowser (Chrome/Firefox) 7 years ago
AlexeyAB baf3fcb491 Calculation TP, FP, FN, avg_IoU for default threshold during mAP calculation 7 years ago
AlexeyAB 65bff2683b It takes into account the Difficult for calculating mAP for PascalVOC 7 years ago
AlexeyAB a1af57d8d6 Added C implementation of calculation mAP (mean average precision) using Darknet 7 years ago
AlexeyAB 6af2cf692c resnet152.cfg 7 years ago
AlexeyAB 2c29218e4e Added compute_mAP.cmd for calculation mAP for Pascal VOC 2007 dataset. 7 years ago
AlexeyAB 9c847647a1 Another compile fix error 7 years ago
AlexeyAB 8df8b36f31 Compile fix 7 years ago
AlexeyAB 14db8f3384 Added linear extrapolation of coordinates 7 years ago