Self-training Room Layout Estimation via Geometry-aware Ray-casting

TL;DR

This paper presents a geometry-aware self-training method for room layout estimation that leverages ray-casting to generate reliable pseudo-labels from unlabeled data, improving accuracy in complex scenes without manual annotations.

Contribution

The proposed framework introduces a novel ray-casting based pseudo-labeling approach that enforces multi-view consistency and handles complex geometries without prior assumptions.

Findings

Significant improvement over state-of-the-art models.

Effective handling of occlusions and complex room shapes.

No reliance on human annotations or assumptions like Manhattan World.

Abstract

In this paper, we introduce a novel geometry-aware self-training framework for room layout estimation models on unseen scenes with unlabeled data. Our approach utilizes a ray-casting formulation to aggregate multiple estimates from different viewing positions, enabling the computation of reliable pseudo-labels for self-training. In particular, our ray-casting approach enforces multi-view consistency along all ray directions and prioritizes spatial proximity to the camera view for geometry reasoning. As a result, our geometry-aware pseudo-labels effectively handle complex room geometries and occluded walls without relying on assumptions such as Manhattan World or planar room walls. Evaluation on publicly available datasets, including synthetic and real-world scenarios, demonstrates significant improvements in current state-of-the-art layout models without using any human annotation.