Open eudemkeon opened 6 years ago
@eudemkeon images are in this format n01514668_19 where first element 'n01514668' is the ID of the class you have in the image. Search for a list containing all 1000 labels with corresponding class ID . Something like https://gist.github.com/aaronpolhamus/964a4411c0906315deb9f4a3723aac57
So just use a python code and split image names and compare and label them. For bounding boxes , you need annotation folder inside imagenet dataset.
Thank you! It really helped me a lot. Thank you for your kindness
Happy to know !
Regards Ahsan
On Wed, Mar 14, 2018 at 11:31 AM, eudemkeon notifications@github.com wrote:
Thank you! It really helped me a lot.
— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/pjreddie/darknet/issues/530#issuecomment-372918623, or mute the thread https://github.com/notifications/unsubscribe-auth/AK9zjlupTHvEtQaw2Ii8r1y5Go-SeNTXks5teLkygaJpZM4Sp3eD .
Hi.
I want to test cfg file and pre-trained weight
from https://pjreddie.com/darknet/imagenet/
such as resnet50, vgg-16 , etc.
Also, I want to train these myself.
The necessary Imagenet2012 datatsets have already been received.
But I can not find label list that available for it.
Could someone provide a label list for the imagenet 2012, 1000 class?
Thanks.