Possible release date

[pourfard]

Hi, thanks for the great article May I ask you what time are you going to release the source code? Will you release the pretrained models too?

[lihengtao]

Hi, thanks for your interest. We are preparing for the open source and will release both the code and models. Please stay tuned.:)

[tensorboy]

Good work, looking forward to the code and pre-trained models.

[ha1ha2hahaha]

Hi，thanks for your great work! when I used the DINOv2 as a backbone, it would give a 1024 embedding as a ouput.But you've written that it will give a tensor with the size of hwc. Could you tell me why?

[lihengtao]

Hi，thanks for your great work! when I used the DINOv2 as a backbone, it would give a 1024 embedding as a ouput.But you've written that it will give a tensor with the size of h_w_c. Could you tell me why?

Hi, you can call DinoVisionTransformer.forward_features to get a dict which contains class token and patch tokens. You can refer https://github.com/facebookresearch/dinov2/blob/main/dinov2/models/vision_transformer.py#LL233C22-L233C22 to get more details. :)

[leolle]

Hi, thanks for your interest. We are preparing for the open source and will release both the code and models. Please stay tuned.:)

加油！

[tieguanyin803]

thanks for the great article

[evolu8]

Great work. Looking forward to the release.

[Jingxi245]

Hello, could you provide the codes and pre-trained models for testing? I want to test using my own data sets.

[NielsRogge]

For those waiting for the release, check out PerSAM (personalized SAM) which is already open-source, I made 2 tutorial notebooks for it here: https://github.com/NielsRogge/Transformers-Tutorials/tree/master/PerSAM

The Matcher authors do compare against PerSAM in their paper

[SSUHan]

Any update? I'm also looking forward to the release!

[zzzyzh]

Please tell me there's a relevant announcement date now, really looking forward to it!

[feivelliu]

When will it be updated?

[simoneangarano]

When are you publishing the code?

[yangliu96]

The code has been released. Thank you for your attention!