Multi-label Plan for DINO-X

[Leekh951]

Currently, DINO-X outputs one bounding box, one label, and one score per detection.
Do you have any plans to enable the model to output multiple possible labels and scores for a single detected object?

[rentainhe]

Hello @Leekh951 , you can detect multiple objects by combining different descriptions with "." in your text prompt.

Here is an example, if you want to detect "person", "dog", and "cat", you can specify your text prompt as "person . dog . cat" and run the inference API with DINO-X.