You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
It would be an excellent addition to have open-set object detectors such as Grounding Dino and/or YoloWorld which are capable of outputting bounding boxes on an image based on a user prompt. This would be amazing alongside MATLAB's implementation of SegmentAnything which already exists.
It would be an excellent addition to have open-set object detectors such as Grounding Dino and/or YoloWorld which are capable of outputting bounding boxes on an image based on a user prompt. This would be amazing alongside MATLAB's implementation of SegmentAnything which already exists.
YoloWorld: https://docs.ultralytics.com/models/yolo-world/
Grounding Dino: https://github.com/IDEA-Research/GroundingDINO
The text was updated successfully, but these errors were encountered: