3D / Point Cloud
Objaverse
800K+ annotated 3D models for training 3D and multimodal models.
100K–1M·CC-BY·Point Cloud (LAS/PLY)
Panoptic Waymo Dataset Summary Panoptic Waymo is a high-resolution LiDAR panoptic segmentation benchmark derived from the Waymo Open Dataset. It provides panoptic annotations for the original Waymo Open Dataset training and validation splits, covering 798 training scenes and 202 validation scenes with 15 stuff classes and 6 thing classes. The benchmark is designed for fine-grained 3D scene understanding and multi-modal perception, leveraging Waymo’s dense… See the full description on the dataset page:
Source: Hugging Face Hub (mohangrim/panoptic-waymo). Metadata imported from the dataset’s Hub tags.