-
Hello team, out of curiosity, from the text I'm imagining that you used the imagenet-21k to train the models, is that right? Or was it the imagenet-1k version?
Thanks!
-
can you provide the imagenet 21k val_map.txt and imagenet1k val_map.txt, thanks!
-
Hello,
I wonder whether the kernel and bias are all set zeros in the released weights of the ViT-B/16 model which has been pre-trained on ImageNet-21k?
-
can I get the dataset you have used? where should I have to implement it?
-
I guess a few more challenging datasets will be popular in the next coming years. Such as meta-dataset, Imagenet-21k. Can you plz add them?
-
Hi,
Thank you for the great work. I wanted to know which ImageNet-21K dataset did you use? Is it the Fall-11 version or Winter-21 version? Thank you
-
Hi all, and thank you for the amazing work!
Is there any chance you could share the weights/checkpoints of DINOv2 ViT-L (Table 1 in the paper) or ViT-H (used for Self-supervised image retrieval in …
-
Does this model also use the original BGR mean and scale values of googlenet?
`mean = [104.00698793 116.66876762 122.67891434]` and `scale = 1.0` ?
In [imagenet-21k-inception.md](https://github.co…
-
Hi, thanks for your great project, it is really great !!
But I met a problem when using your tool to download ImageNet-21k(https://image-net.org/data/winter21_whole.tar.gz). It is a single file an…
-
Thanks for your great work!
I have a question when using your ResNet50 model as pretrained weights of Faster R-CNN in Detectron2: your 21K pretrained weights gives 8 point lower mAP than MSRA 1K pret…