search

JasonQSY / 3DOI

[ICCV 2023] Understanding 3D Object Interaction from a Single Image
37 stars 2 forks source link

Questions about the image annotations #3

Open TeleeMa opened 1 month ago

TeleeMa commented 1 month ago

Hi,

Thank you for your great work. Are there any text annotations of the 3DOI images, like describing the annotated objects and actions?

JasonQSY commented 1 month ago

Thanks for your interests! There are only limited action annotations (pull, push and others). And it is not open-vocabulary. For objects, we mainly include refrigerators, microwave, door, drawer, cabinet etc. and it might be possible to get good-quality pseudo annotations with a good object detector.