Feature request: YOLO-World/Grounding DINO (Zero shot object detection)

xenova / transformers.js

State-of-the-art Machine Learning for the web. Run 🤗 Transformers directly in your browser, with no need for a server!

https://huggingface.co/docs/transformers.js

Apache License 2.0

9.71k stars 571 forks source link

Feature request: YOLO-World/Grounding DINO (Zero shot object detection) #792

Open rogueturnip opened 3 weeks ago

rogueturnip commented 3 weeks ago

Question

Hi!

I'm trying out some of the zero shot capabilities and I've been working with the owlv2 but I was wondering, is support for yolo-world and grounding Dino coming? They seem to be faster than owlv2.

Thanks!

xenova commented 3 weeks ago

Grounding DINO was added to transformers ~2 months ago in https://github.com/huggingface/transformers/pull/26087, so it's definitely possible! Maybe a contributor would like to get the ONNX conversion working in Optimum - that would unblock us and allow us to add it to Transformers.js

rogueturnip commented 3 weeks ago

Thanks for that! Looks like the feature request for Optimum was made last month https://github.com/huggingface/optimum/issues/1849

pacozaa commented 2 days ago

Following this one!