SynthCam3D: Semantic Understanding With Synthetic Indoor Scenes

TL;DR

SynthCam3D introduces a deep autoencoder trained solely on synthetic indoor scene data for real-time semantic segmentation, enabling effective scene understanding without noise modeling.

Contribution

The paper presents a novel synthetic 3D scene library and a deep autoencoder approach for real-time semantic segmentation of indoor scenes from depth data.

Findings

Effective segmentation of real scenes without noise modeling

Successful training on synthetic data for real-world application

Preliminary results show promising scene understanding capabilities

Abstract

We are interested in automatic scene understanding from geometric cues. To this end, we aim to bring semantic segmentation in the loop of real-time reconstruction. Our semantic segmentation is built on a deep autoencoder stack trained exclusively on synthetic depth data generated from our novel 3D scene library, SynthCam3D. Importantly, our network is able to segment real world scenes without any noise modelling. We present encouraging preliminary results.