Transferable Semi-supervised 3D Object Detection from RGB-D Data

We investigate the direction of training a 3D object detector for new object classes from only 2D bounding box labels of these new classes, while simultaneously transferring information from 3D bounding box labels of the existing classes. To this end, we propose a transferable semi-supervised 3D obj...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:arXiv.org 2019-04
Hauptverfasser: Tang, Yew Siang, Lee, Gim Hee
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page
container_issue
container_start_page
container_title arXiv.org
container_volume
creator Tang, Yew Siang
Lee, Gim Hee
description We investigate the direction of training a 3D object detector for new object classes from only 2D bounding box labels of these new classes, while simultaneously transferring information from 3D bounding box labels of the existing classes. To this end, we propose a transferable semi-supervised 3D object detection model that learns a 3D object detector network from training data with two disjoint sets of object classes - a set of strong classes with both 2D and 3D box labels, and another set of weak classes with only 2D box labels. In particular, we suggest a relaxed reprojection loss, box prior loss and a Box-to-Point Cloud Fit network that allow us to effectively transfer useful 3D information from the strong classes to the weak classes during training, and consequently, enable the network to detect 3D objects in the weak classes during inference. Experimental results show that our proposed algorithm outperforms baseline approaches and achieves promising results compared to fully-supervised approaches on the SUN-RGBD and KITTI datasets. Furthermore, we show that our Box-to-Point Cloud Fit network improves performances of the fully-supervised approaches on both datasets.
format Article
fullrecord <record><control><sourceid>proquest</sourceid><recordid>TN_cdi_proquest_journals_2213628660</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2213628660</sourcerecordid><originalsourceid>FETCH-proquest_journals_22136286603</originalsourceid><addsrcrecordid>eNqNyr0OgjAUQOHGxESivMNNnJuUW6iMRuvPZqLspOglgQDFtvj8MvgATt9wzoJFKGXC8xRxxWLvWyEEqh1mmYzYvnBm8DU5U3UED-ob7qeR3Kfx9AKp4Va19AygKcw0doDa2R7ulwPXoE0wG7asTecp_rlm2_OpOF756Ox7Ih_K1k5umFOJmEiFuVJC_nd9AbQ_N7I</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>2213628660</pqid></control><display><type>article</type><title>Transferable Semi-supervised 3D Object Detection from RGB-D Data</title><source>Free E- Journals</source><creator>Tang, Yew Siang ; Lee, Gim Hee</creator><creatorcontrib>Tang, Yew Siang ; Lee, Gim Hee</creatorcontrib><description>We investigate the direction of training a 3D object detector for new object classes from only 2D bounding box labels of these new classes, while simultaneously transferring information from 3D bounding box labels of the existing classes. To this end, we propose a transferable semi-supervised 3D object detection model that learns a 3D object detector network from training data with two disjoint sets of object classes - a set of strong classes with both 2D and 3D box labels, and another set of weak classes with only 2D box labels. In particular, we suggest a relaxed reprojection loss, box prior loss and a Box-to-Point Cloud Fit network that allow us to effectively transfer useful 3D information from the strong classes to the weak classes during training, and consequently, enable the network to detect 3D objects in the weak classes during inference. Experimental results show that our proposed algorithm outperforms baseline approaches and achieves promising results compared to fully-supervised approaches on the SUN-RGBD and KITTI datasets. Furthermore, we show that our Box-to-Point Cloud Fit network improves performances of the fully-supervised approaches on both datasets.</description><identifier>EISSN: 2331-8422</identifier><language>eng</language><publisher>Ithaca: Cornell University Library, arXiv.org</publisher><subject>Algorithms ; Datasets ; Labels ; Object recognition ; Performance enhancement ; Three dimensional models ; Training</subject><ispartof>arXiv.org, 2019-04</ispartof><rights>2019. This work is published under http://arxiv.org/licenses/nonexclusive-distrib/1.0/ (the “License”). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.</rights><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>780,784</link.rule.ids></links><search><creatorcontrib>Tang, Yew Siang</creatorcontrib><creatorcontrib>Lee, Gim Hee</creatorcontrib><title>Transferable Semi-supervised 3D Object Detection from RGB-D Data</title><title>arXiv.org</title><description>We investigate the direction of training a 3D object detector for new object classes from only 2D bounding box labels of these new classes, while simultaneously transferring information from 3D bounding box labels of the existing classes. To this end, we propose a transferable semi-supervised 3D object detection model that learns a 3D object detector network from training data with two disjoint sets of object classes - a set of strong classes with both 2D and 3D box labels, and another set of weak classes with only 2D box labels. In particular, we suggest a relaxed reprojection loss, box prior loss and a Box-to-Point Cloud Fit network that allow us to effectively transfer useful 3D information from the strong classes to the weak classes during training, and consequently, enable the network to detect 3D objects in the weak classes during inference. Experimental results show that our proposed algorithm outperforms baseline approaches and achieves promising results compared to fully-supervised approaches on the SUN-RGBD and KITTI datasets. Furthermore, we show that our Box-to-Point Cloud Fit network improves performances of the fully-supervised approaches on both datasets.</description><subject>Algorithms</subject><subject>Datasets</subject><subject>Labels</subject><subject>Object recognition</subject><subject>Performance enhancement</subject><subject>Three dimensional models</subject><subject>Training</subject><issn>2331-8422</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2019</creationdate><recordtype>article</recordtype><sourceid>ABUWG</sourceid><sourceid>AFKRA</sourceid><sourceid>AZQEC</sourceid><sourceid>BENPR</sourceid><sourceid>CCPQU</sourceid><sourceid>DWQXO</sourceid><recordid>eNqNyr0OgjAUQOHGxESivMNNnJuUW6iMRuvPZqLspOglgQDFtvj8MvgATt9wzoJFKGXC8xRxxWLvWyEEqh1mmYzYvnBm8DU5U3UED-ob7qeR3Kfx9AKp4Va19AygKcw0doDa2R7ulwPXoE0wG7asTecp_rlm2_OpOF756Ox7Ih_K1k5umFOJmEiFuVJC_nd9AbQ_N7I</recordid><startdate>20190423</startdate><enddate>20190423</enddate><creator>Tang, Yew Siang</creator><creator>Lee, Gim Hee</creator><general>Cornell University Library, arXiv.org</general><scope>8FE</scope><scope>8FG</scope><scope>ABJCF</scope><scope>ABUWG</scope><scope>AFKRA</scope><scope>AZQEC</scope><scope>BENPR</scope><scope>BGLVJ</scope><scope>CCPQU</scope><scope>DWQXO</scope><scope>HCIFZ</scope><scope>L6V</scope><scope>M7S</scope><scope>PIMPY</scope><scope>PQEST</scope><scope>PQQKQ</scope><scope>PQUKI</scope><scope>PRINS</scope><scope>PTHSS</scope></search><sort><creationdate>20190423</creationdate><title>Transferable Semi-supervised 3D Object Detection from RGB-D Data</title><author>Tang, Yew Siang ; Lee, Gim Hee</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-proquest_journals_22136286603</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2019</creationdate><topic>Algorithms</topic><topic>Datasets</topic><topic>Labels</topic><topic>Object recognition</topic><topic>Performance enhancement</topic><topic>Three dimensional models</topic><topic>Training</topic><toplevel>online_resources</toplevel><creatorcontrib>Tang, Yew Siang</creatorcontrib><creatorcontrib>Lee, Gim Hee</creatorcontrib><collection>ProQuest SciTech Collection</collection><collection>ProQuest Technology Collection</collection><collection>Materials Science &amp; Engineering Collection</collection><collection>ProQuest Central (Alumni Edition)</collection><collection>ProQuest Central UK/Ireland</collection><collection>ProQuest Central Essentials</collection><collection>ProQuest Central</collection><collection>Technology Collection</collection><collection>ProQuest One Community College</collection><collection>ProQuest Central Korea</collection><collection>SciTech Premium Collection</collection><collection>ProQuest Engineering Collection</collection><collection>Engineering Database</collection><collection>Publicly Available Content Database</collection><collection>ProQuest One Academic Eastern Edition (DO NOT USE)</collection><collection>ProQuest One Academic</collection><collection>ProQuest One Academic UKI Edition</collection><collection>ProQuest Central China</collection><collection>Engineering Collection</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Tang, Yew Siang</au><au>Lee, Gim Hee</au><format>book</format><genre>document</genre><ristype>GEN</ristype><atitle>Transferable Semi-supervised 3D Object Detection from RGB-D Data</atitle><jtitle>arXiv.org</jtitle><date>2019-04-23</date><risdate>2019</risdate><eissn>2331-8422</eissn><abstract>We investigate the direction of training a 3D object detector for new object classes from only 2D bounding box labels of these new classes, while simultaneously transferring information from 3D bounding box labels of the existing classes. To this end, we propose a transferable semi-supervised 3D object detection model that learns a 3D object detector network from training data with two disjoint sets of object classes - a set of strong classes with both 2D and 3D box labels, and another set of weak classes with only 2D box labels. In particular, we suggest a relaxed reprojection loss, box prior loss and a Box-to-Point Cloud Fit network that allow us to effectively transfer useful 3D information from the strong classes to the weak classes during training, and consequently, enable the network to detect 3D objects in the weak classes during inference. Experimental results show that our proposed algorithm outperforms baseline approaches and achieves promising results compared to fully-supervised approaches on the SUN-RGBD and KITTI datasets. Furthermore, we show that our Box-to-Point Cloud Fit network improves performances of the fully-supervised approaches on both datasets.</abstract><cop>Ithaca</cop><pub>Cornell University Library, arXiv.org</pub><oa>free_for_read</oa></addata></record>
fulltext fulltext
identifier EISSN: 2331-8422
ispartof arXiv.org, 2019-04
issn 2331-8422
language eng
recordid cdi_proquest_journals_2213628660
source Free E- Journals
subjects Algorithms
Datasets
Labels
Object recognition
Performance enhancement
Three dimensional models
Training
title Transferable Semi-supervised 3D Object Detection from RGB-D Data
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2024-12-19T12%3A15%3A05IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-proquest&rft_val_fmt=info:ofi/fmt:kev:mtx:book&rft.genre=document&rft.atitle=Transferable%20Semi-supervised%203D%20Object%20Detection%20from%20RGB-D%20Data&rft.jtitle=arXiv.org&rft.au=Tang,%20Yew%20Siang&rft.date=2019-04-23&rft.eissn=2331-8422&rft_id=info:doi/&rft_dat=%3Cproquest%3E2213628660%3C/proquest%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_pqid=2213628660&rft_id=info:pmid/&rfr_iscdi=true