YouTube-Occ: Learning Indoor 3D Semantic Occupancy Prediction from YouTube Videos

TL;DR

This paper introduces YouTube-Occ, a self-supervised method that learns 3D indoor semantic occupancy prediction from YouTube videos without requiring detailed camera parameters, achieving state-of-the-art zero-shot results.

Contribution

It presents a novel approach leveraging web-sourced indoor videos and vision foundation models for 3D perception without geometric annotations.

Findings

Achieves state-of-the-art zero-shot performance on NYUv2 and OccScanNet.

Demonstrates effective 3D indoor understanding using only internet videos.

Eliminates the need for precise camera calibration in 3D learning.

Abstract

3D semantic occupancy prediction in the past was considered to require precise geometric relationships in order to enable effective training. However, in complex indoor environments, the large-scale and widespread collection of data, along with the necessity for fine-grained annotations, becomes impractical due to the complexity of data acquisition setups and privacy concerns. In this paper, we demonstrate that 3D spatially-accurate training can be achieved using only indoor Internet data, without the need for any pre-knowledge of intrinsic or extrinsic camera parameters. In our framework, we collect a web dataset, YouTube-Occ, which comprises house tour videos from YouTube, providing abundant real house scenes for 3D representation learning. Upon on this web dataset, we establish a fully self-supervised model to leverage accessible 2D prior knowledge for reaching powerful 3D indoor perception. Specifically, we harness the advantages of the prosperous vision foundation models, distilling the 2D region-level knowledge into the occupancy network by grouping the similar pixels into superpixels. Experimental results show that our method achieves state-of-the-art zero-shot performance on two popular benchmarks (NYUv2 and OccScanNet