Language-based object detectors (GLIP, Grounding DINO) support for MMDetection
Hello ! I've added support of language-based object detectors (GLIP, Grounding DINO) from the latest release of MMdetection which require a text prompt in order to work. I've also included a test and download utils.
Users can provide a custom text prompt using --text argument. If none is provided the class names are used as the text prompt.
Amazing PR @AMauri996 ! 🔥 🚀
Can you also include a quickstart notebook as in other integrations so that users know how to use it?
You can take a look at other notebook files here: https://github.com/obss/sahi/tree/main/demo
Hello @fcakyon ! Thank you for your review. I've added the demo for GLIP here.
@AMauri996 can you fix the styling?
Is there something missing? Happy to jump and help in case it is needed
@fcakyon I fixed the styling, it should be good now
@AMauri996 thanks a lot!
Please fix the nltk based failure in the continuous integration tests @AMauri996 @RobertoPalomares 👍🏻
https://github.com/obss/sahi/pull/1003/checks
@fcakyon I fixed the nltk error in the tests
Thanks for the fast fix, @AMauri996. Can you please update demo notebook using the latest up-to-date mmdetection notebook available in the sahi repo? Then we would be glad to merge your contribution!
Hello @fcakyon ! I've solved the merge conflict caused by the mmdetection notebook
Hello @fcakyon ! I've solved the merge conflict caused by the mmdetection notebook
Thanks for your efforts @AMauri996 !
Why do you think the mmdet tests are failing? 🤔 https://github.com/obss/sahi/actions/runs/10006492987/job/28194626143?pr=1003#step:19:1340