IDEA-Research / DINO-X-API

DINO-X: The World's Top-Performing Vision Model for Open-World Object Detection and Understanding
https://arxiv.org/abs/2411.14347
Apache License 2.0
368 stars 16 forks source link

very need for demo on Segmentation Phrase Grounding #3

Open arlisgreat opened 4 days ago

rentainhe commented 4 days ago

Hello, you can use some open-sourced language tools to get the noun chunks from the sentence, like spacy, then you can concatenate them using . and use DINO-X to predict their corresponding boxes and masks, we will offer some examples in the future release.

rentainhe commented 4 days ago

There are still some errors in our prediction mask API, we will fix them later and release the segmentation demo, and we are still enhancing the capabilities of the mask head.

arlisgreat commented 4 days ago

@rentainhe great thanks to you!!!!! we need it for our coureswork,so when will it be released?

rentainhe commented 4 days ago

We will release it within this week

xyb1314 commented 3 days ago

Waiting for prediction mask API, too! Find hard to parse the mask=ObjectMask params in the response !