ImOV3D: LearningOpen-VocabularyPointClouds 3DObjectDetectionfromOnly2DImages

Neural Information Processing Systems 

Open-vocabulary 3D object detection (OV-3Det) aims to generalize beyond the limited number ofbasecategories labeled during thetraining phase. Thebiggest bottleneck is the scarcity of annotated 3D data, whereas 2D image datasets are abundantandrichlyannotated.

Similar Docs  Excel Report  more

TitleSimilaritySource
None found