FlowSSC: Universal Generative Monocular Semantic Scene Completion via One-Step Latent Diffusion

TL;DR

FlowSSC introduces a one-step latent diffusion framework for monocular semantic scene completion, significantly improving detail generation and spatial consistency in 3D reconstructions from single images, suitable for real-time applications.

Contribution

The paper presents FlowSSC, a novel generative approach that integrates with existing methods, enabling fast, high-quality 3D scene completion from monocular images using a shortcut diffusion mechanism.

Findings

Achieves state-of-the-art results on SemanticKITTI

Operates in a single diffusion step for real-time inference

Outperforms existing baselines in detail and spatial accuracy

Abstract

Semantic Scene Completion (SSC) from monocular RGB images is a fundamental yet challenging task due to the inherent ambiguity of inferring occluded 3D geometry from a single view. While feed-forward methods have made progress, they often struggle to generate plausible details in occluded regions and preserve the fundamental spatial relationships of objects. Such accurate generative reasoning capability for the entire 3D space is critical in real-world applications. In this paper, we present FlowSSC, the first generative framework applied directly to monocular semantic scene completion. FlowSSC treats the SSC task as a conditional generation problem and can seamlessly integrate with existing feed-forward SSC methods to significantly boost their performance. To achieve real-time inference without compromising quality, we introduce Shortcut Flow-matching that operates in a compact triplane latent space. Unlike standard diffusion models that require hundreds of steps, our method utilizes a shortcut mechanism to achieve high-fidelity generation in a single step, enabling practical deployment in autonomous systems. Extensive experiments on SemanticKITTI demonstrate that FlowSSC achieves state-of-the-art performance, significantly outperforming existing baselines.