Data-Efficient Semantic Segmentation of 3D Point Clouds via Open-Vocabulary Image Segmentation-based Pseudo-Labeling

Name: Data-Efficient Semantic Segmentation of 3D Point Clouds via Open-Vocabulary Image Segmentation-based Pseudo-Labeling
Uploaded: 2026-10-23T04:43:22.053489+00:00
Duration: 411 s
Description: PLOVIS uses an Open-Vocabulary Image Segmentation (OVIS) model to create pseudo labels for 3D point cloud segmentation when training data is scarce. It renders point clouds into images, uses OVIS to generate class predictions, and filters these labels rigorously. This method allows for high accuracy even with very few annotated scenes or points.

PLOVIS uses an Open-Vocabulary Image Segmentation (OVIS) model to create pseudo labels for 3D point cloud segmentation when training data is scarce. It renders point clouds into images, uses OVIS to generate class predictions, and filters these labels rigorously. This method allows for high accuracy even with very few annotated scenes or points.

Open the full episode