search

IDEA-Research / GroundingDINO

[ECCV 2024] Official implementation of the paper "Grounding DINO: Marrying DINO with Grounded Pre-Training for Open-Set Object Detection"
https://arxiv.org/abs/2303.05499
Apache License 2.0
6.9k stars 698 forks source link

Can GroundingDINO do one-shot detection? #207

Open nickponline opened 1 year ago

nickponline commented 1 year ago

Can Grounding DINO be used for image conditioned one-shot or few-shot detection like this?

image_cond_wiki_circuits_1

The image is from OWL-VIT which can do be prompted with either text or an image patch. I feel that DINO should be able to do the same thing and likely perform better?

AdamMayor2018 commented 1 year ago

Is the problem solved? Same here.

Nancis1130 commented 1 year ago

same problem.

jonomillin commented 1 year ago

Same question!

AdamMayor2018 commented 1 year ago

same here.

nickponline commented 1 year ago

@rentainhe @SlongLiu perhaps you know the answer here and could point us in the right direction?

mikewinn1 commented 1 year ago

same here

Alaric423 commented 9 months ago

Any update here?

YoungSean commented 6 months ago

Please check out NIDS-Net. It includes Grounding-DINO to handle these tasks. It can detect things with just one or a few template images. Moreover, it may not need any training.