Semantic Scene Completion from a Single 360-Degree Image and Depth Map

Aloisio Dourado, Hansung Kim, Teofilo E. de Campos, Adrian Hilton

Abstract

We present a method for Semantic Scene Completion (SSC) of complete indoor scenes from a single 360â—¦ RGB image and corresponding depth map using a Deep Convolution Neural Network that takes advantage of existing datasets of synthetic and real RGB-D images for training. Recent works on SSC only perform occupancy prediction of small regions of the room covered by the field-of-view of the sensor in use, which implies the need of multiple images to cover the whole scene, being an inappropriate method for dynamic scenes. Our approach uses only a single 360â—¦ image with its corresponding depth map to infer the occupancy and semantic labels of the whole room. Using one single image is important to allow predictions with no previous knowledge of the scene and enable extension to dynamic scene applications. We evaluated our method on two 360â—¦ image datasets: a high-quality 360â—¦ RGB-D dataset gathered with a Matterport sensor and low-quality 360â—¦ RGB-D images generated with a pair of commercial 360â—¦ cameras and stereo matching. The experiments showed that the proposed pipeline performs SSC not only with Matterport cameras but also with more affordable 360â—¦ cameras, which adds a great number of potential applications, including immersive spatial audio reproduction, augmented reality, assistive computing and robotics.

Download


Paper Citation


in Harvard Style

Dourado A., Kim H., E. de Campos T. and Hilton A. (2020). Semantic Scene Completion from a Single 360-Degree Image and Depth Map.In Proceedings of the 15th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications - Volume 5: VISAPP, ISBN 978-989-758-402-2, pages 36-46. DOI: 10.5220/0008877700360046


in Bibtex Style

@conference{visapp20,
author={Aloisio Dourado and Hansung Kim and Teofilo E. de Campos and Adrian Hilton},
title={Semantic Scene Completion from a Single 360-Degree Image and Depth Map},
booktitle={Proceedings of the 15th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications - Volume 5: VISAPP,},
year={2020},
pages={36-46},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0008877700360046},
isbn={978-989-758-402-2},
}


in EndNote Style

TY - CONF

JO - Proceedings of the 15th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications - Volume 5: VISAPP,
TI - Semantic Scene Completion from a Single 360-Degree Image and Depth Map
SN - 978-989-758-402-2
AU - Dourado A.
AU - Kim H.
AU - E. de Campos T.
AU - Hilton A.
PY - 2020
SP - 36
EP - 46
DO - 10.5220/0008877700360046