AlexeyAB / darknet

YOLOv4 / Scaled-YOLOv4 / YOLO - Neural Networks for Object Detection (Windows and Linux version of Darknet )
http://pjreddie.com/darknet/
Other
21.75k stars 7.96k forks source link

what is difference between yolov4.conv137 and csdarknet53-omega.conv.105 #5804

Open RoyLeeDongHee opened 4 years ago

RoyLeeDongHee commented 4 years ago

hi. i'll start to train yolov4 with MSCOCO dataset.

what is difference between yolov4.conv137 and csdarknet53-omega.conv.105 ?

why you suggest csdarknet53-omega.conv.105 to train with MSCOCO?

image

THX.

WongKinYiu commented 4 years ago

yolov4.conv137 is coco pretrained model. csdarknet53-omega.conv.105 is imagenet pretrained model.

you won't have coco pretrained model before training coco, so csdarknet53-omega.conv.105 is suggested to train the coco model. and if your dataset is larger than coco, for example, openimage, i will also suggest you to use csdarknet53-omega.conv.105.

ZhixinLai commented 3 years ago

hi, where can I download csdarknet53-omega.conv.105? @WongKinYiu @AlexeyAB @RoyLeeDongHee Thanks for help~

AlexeyAB commented 3 years ago

https://github.com/AlexeyAB/darknet/releases/download/darknet_yolo_v4_pre/csdarknet53-omega.conv.105

ZhixinLai commented 3 years ago

https://github.com/AlexeyAB/darknet/releases/download/darknet_yolo_v4_pre/csdarknet53-omega.conv.105

thanks