An Analysis of 3D Indoor Scene Segmentation Based on Images, Point Cloud and Voxel Data

碩士 === 國立臺灣大學 === 資訊工程學研究所 === 107 === The deep learning technology has brought great success in image classification, object detection and semantic segmentation tasks. Recent years, the advent of inexpensive depth sensors hugely motivate 3D research area and real scene reconstruction datasets such a...

Full description

Bibliographic Details
Main Authors: Hung-Yueh Chiang, 江泓樂
Other Authors: Winston Hsu
Format: Others
Language:en_US
Published: 2018
Online Access:http://ndltd.ncl.edu.tw/handle/pvbj47
id ndltd-TW-107NTU05392017
record_format oai_dc
spelling ndltd-TW-107NTU053920172019-06-27T05:48:11Z http://ndltd.ncl.edu.tw/handle/pvbj47 An Analysis of 3D Indoor Scene Segmentation Based on Images, Point Cloud and Voxel Data 基於影像、點雲與立素之3D室內環境語意切割分析與討論 Hung-Yueh Chiang 江泓樂 碩士 國立臺灣大學 資訊工程學研究所 107 The deep learning technology has brought great success in image classification, object detection and semantic segmentation tasks. Recent years, the advent of inexpensive depth sensors hugely motivate 3D research area and real scene reconstruction datasets such as ScanNet [5] and Matterport3D [1] have been proposed. However, the problem of 3D scene semantic segmentation still remains new and challenging due to many variance of 3D data type (e.g. image, voxel, point cloud). Other difficulties such as suffering from high computation cost and the scarcity of data dispel the research progress of 3D segmentation. In this paper, we study 3D indoor scene segmentation problem with three different types of 3D data, which we categorize into image-based, voxel-based and point-based. We experiment on different input signals (e.g. color, depth, normal) and verify their effectiveness and performance in different data type networks. We further study fusion methods and improve the performance by using off-the-shelf deep models and by leveraging data modalities in the paper. Winston Hsu 徐宏民 2018 學位論文 ; thesis 31 en_US
collection NDLTD
language en_US
format Others
sources NDLTD
description 碩士 === 國立臺灣大學 === 資訊工程學研究所 === 107 === The deep learning technology has brought great success in image classification, object detection and semantic segmentation tasks. Recent years, the advent of inexpensive depth sensors hugely motivate 3D research area and real scene reconstruction datasets such as ScanNet [5] and Matterport3D [1] have been proposed. However, the problem of 3D scene semantic segmentation still remains new and challenging due to many variance of 3D data type (e.g. image, voxel, point cloud). Other difficulties such as suffering from high computation cost and the scarcity of data dispel the research progress of 3D segmentation. In this paper, we study 3D indoor scene segmentation problem with three different types of 3D data, which we categorize into image-based, voxel-based and point-based. We experiment on different input signals (e.g. color, depth, normal) and verify their effectiveness and performance in different data type networks. We further study fusion methods and improve the performance by using off-the-shelf deep models and by leveraging data modalities in the paper.
author2 Winston Hsu
author_facet Winston Hsu
Hung-Yueh Chiang
江泓樂
author Hung-Yueh Chiang
江泓樂
spellingShingle Hung-Yueh Chiang
江泓樂
An Analysis of 3D Indoor Scene Segmentation Based on Images, Point Cloud and Voxel Data
author_sort Hung-Yueh Chiang
title An Analysis of 3D Indoor Scene Segmentation Based on Images, Point Cloud and Voxel Data
title_short An Analysis of 3D Indoor Scene Segmentation Based on Images, Point Cloud and Voxel Data
title_full An Analysis of 3D Indoor Scene Segmentation Based on Images, Point Cloud and Voxel Data
title_fullStr An Analysis of 3D Indoor Scene Segmentation Based on Images, Point Cloud and Voxel Data
title_full_unstemmed An Analysis of 3D Indoor Scene Segmentation Based on Images, Point Cloud and Voxel Data
title_sort analysis of 3d indoor scene segmentation based on images, point cloud and voxel data
publishDate 2018
url http://ndltd.ncl.edu.tw/handle/pvbj47
work_keys_str_mv AT hungyuehchiang ananalysisof3dindoorscenesegmentationbasedonimagespointcloudandvoxeldata
AT jiānghónglè ananalysisof3dindoorscenesegmentationbasedonimagespointcloudandvoxeldata
AT hungyuehchiang jīyúyǐngxiàngdiǎnyúnyǔlìsùzhī3dshìnèihuánjìngyǔyìqiègēfēnxīyǔtǎolùn
AT jiānghónglè jīyúyǐngxiàngdiǎnyúnyǔlìsùzhī3dshìnèihuánjìngyǔyìqiègēfēnxīyǔtǎolùn
AT hungyuehchiang analysisof3dindoorscenesegmentationbasedonimagespointcloudandvoxeldata
AT jiānghónglè analysisof3dindoorscenesegmentationbasedonimagespointcloudandvoxeldata
_version_ 1719214042522845184