search

facebookresearch / ov-seg

This is the official PyTorch implementation of the paper Open-Vocabulary Semantic Segmentation with Mask-adapted CLIP.
Other
689 stars 61 forks source link

create ONNX for anylabeling #11

Closed aiyou9 closed 1 year ago

aiyou9 commented 1 year ago

I want import OV-seg to anylabeling, please give ONNX format or some advise for me, thanks a lot ! https://github.com/vietanhdev/anylabeling

Jeff-LiangF commented 1 year ago

Hi @aiyou9 ,

Thanks for your interest! Unfortunately, we don't have an ONNX format and don't have a plan either. You are very welcome to do it by yourself and we would definitely help you to do so.

I noticed in anylabeling you have an implementation of SAM. We incorporate OVSeg CLIP w/ SAM, enabling SAM’s text prompts, try our demo in hf ovseg. Regarding the implementation, I believe it should be easy since you only need to convert the OVSeg CLIP to the ONNX format. Our OVSeg CLIP has the exact format as OpenAI's CLIP, so you may easily find the tutorials to convert CLIP to ONNX.

Feel free to reopen this issue or directly reach out to me if you have any questions. Thanks!