search

leondgarse / keras_cv_attention_models

Keras beit,caformer,CMT,CoAtNet,convnext,davit,dino,efficientdet,edgenext,efficientformer,efficientnet,eva,fasternet,fastervit,fastvit,flexivit,gcvit,ghostnet,gpvit,hornet,hiera,iformer,inceptionnext,lcnet,levit,maxvit,mobilevit,moganet,nat,nfnets,pvt,swin,tinynet,tinyvit,uniformer,volo,vanillanet,yolor,yolov7,yolov8,yolox,gpt2,llama2, alias kecam
MIT License
595 stars 95 forks source link

DINO v2 backbone #116

Closed anshkumar closed 1 year ago

anshkumar commented 1 year ago

Would like to add DINOv2 backbone.

paper: https://arxiv.org/abs/2304.07193 github: https://github.com/facebookresearch/dinov2

leondgarse commented 1 year ago

Ok, added in the TODO list.

leondgarse commented 1 year ago

DINOv2 models are added in keras_cv_attention_models/beit.

anshkumar commented 1 year ago

That's great. You should also add link to https://paperswithcode.com/paper/dinov2-learning-robust-visual-features. Also, for other implementations too.

leondgarse commented 1 year ago

I'm hesitate adding the link, same situation like EVA models. As this is only model architecture and weights, and the architecture is just a ViT, but it's the training methods being the key point of those papers...

anshkumar commented 1 year ago

You are right, but model architecture and weights are only needed in the majority. Most of the links are like this only. Adding the link will add more attraction of researchers to this repo. You should definitely add the link.