darknet

Commit Graph

Author	SHA1	Message	Date
Alexey	11e6f4b081	Merge pull request #492 from sivagnanamn/master Minor changes to keep MAC OS happy	7 years ago
AlexeyAB	8f1f5cbf83	3.5x times accelerated data augmentation using OpenCV for training Yolo on DGX and multi-GPU	7 years ago
sivagnanamn	013031ce3f	Minor changes to keep MAC OSX happy	7 years ago
AlexeyAB	101de2b07a	More stable web-cam	7 years ago
AlexeyAB	fe4f01b1d8	Fixed uses of WebCam for OpenCV 3.x	7 years ago
Alexey	3e5abe0680	Update Readme.md	7 years ago
AlexeyAB	ced198e939	Fixed gpu_id for DLL/SO	7 years ago
AlexeyAB	537d135feb	Improve training performance - batch-norm using cuDNN.	7 years ago
AlexeyAB	2f52cfeb07	Fixed compile error	7 years ago
AlexeyAB	e947bf29af	C wrappers	7 years ago
AlexeyAB	164ba34279	Fixed bug, that save weights-file for each iteration	7 years ago
AlexeyAB	371f21171a	shortcut_layer resize for random=1	7 years ago
AlexeyAB	df076653e0	Minor fix	7 years ago
AlexeyAB	15c89e7a71	Added include for sprintf	7 years ago
AlexeyAB	880cf187d8	Fixed multi-GPU training for Tensor Cores	7 years ago
AlexeyAB	a6c51e3b75	Fixes for small objects. And fixes of densenet201_yolo.cfg, resnet50_yolo.cfg, resnet152_yolo.cfg	7 years ago
AlexeyAB	100d6f7801	Remove 1 additional conv 3x3 kernel layer from resnet152_yolo.cfg and resnet50_yolo.cfg	7 years ago
AlexeyAB	ae1768e583	Removed random=1 from resnet152_yolo.cfg. Until resize_network() isn't supported for [shortcut] layer	7 years ago
AlexeyAB	e90afa6ec1	resnet152_yolo.cfg uses learning_rate, steps, scales, decay, momentum as in the RetinaNet	7 years ago
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. alpha = 0.5, gamma = 2	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

1 2 3 4 5 ...

627 Commits (11e6f4b081588a08687952bd5ddaaa04fece6144) All Branches Search

627 Commits (11e6f4b081588a08687952bd5ddaaa04fece6144)

All Branches