Access the embeddings of detected objects

[AelaLS]

Hello,

First of all, thank you for your work on OneFormer and for making the code publicly available.

I am currently using OneFormer for panoptic segmentation and would like to know if it is possible to access the embeddings of detected objects.

My goal is to use these embeddings for downstream tasks such as similarity comparison or clustering between segmented objects.

I checked the model outputs and the codebase, but I could not find a clear way to retrieve these representations directly.

Is there an official or recommended way to access them?

And if this is not currently supported, do you know of any panoptic segmentation models that expose object-level embeddings in a convenient way?

Thanks!