PP-HumanSeg: Connectivity-Aware Portrait Segmentation with a Large-Scale Teleconferencing Video Dataset

As the COVID-19 pandemic rampages across the world, the demands of video conferencing surge. To this end, real-time portrait segmentation becomes a popular feature to replace backgrounds of conferencing participants. While feature-rich datasets, models and algorithms have been offered for segmentati...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Chu, Lutao, Liu, Yi, Wu, Zewu, Tang, Shiyu, Chen, Guowei, Hao, Yuying, Peng, Juncai, Yu, Zhiliang, Chen, Zeyu, Lai, Baohua, Xiong, Haoyi
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page
container_issue
container_start_page
container_title
container_volume
creator Chu, Lutao
Liu, Yi
Wu, Zewu
Tang, Shiyu
Chen, Guowei
Hao, Yuying
Peng, Juncai
Yu, Zhiliang
Chen, Zeyu
Lai, Baohua
Xiong, Haoyi
description As the COVID-19 pandemic rampages across the world, the demands of video conferencing surge. To this end, real-time portrait segmentation becomes a popular feature to replace backgrounds of conferencing participants. While feature-rich datasets, models and algorithms have been offered for segmentation that extract body postures from life scenes, portrait segmentation has yet not been well covered in a video conferencing context. To facilitate the progress in this field, we introduce an open-source solution named PP-HumanSeg. This work is the first to construct a large-scale video portrait dataset that contains 291 videos from 23 conference scenes with 14K fine-labeled frames and extensions to multi-camera teleconferencing. Furthermore, we propose a novel Semantic Connectivity-aware Learning (SCL) for semantic segmentation, which introduces a semantic connectivity-aware loss to improve the quality of segmentation results from the perspective of connectivity. And we propose an ultra-lightweight model with SCL for practical portrait segmentation, which achieves the best trade-off between IoU and the speed of inference. Extensive evaluations on our dataset demonstrate the superiority of SCL and our model. The source code is available at https://github.com/PaddlePaddle/PaddleSeg.
doi_str_mv 10.48550/arxiv.2112.07146
format Article
fullrecord <record><control><sourceid>arxiv_GOX</sourceid><recordid>TN_cdi_arxiv_primary_2112_07146</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2112_07146</sourcerecordid><originalsourceid>FETCH-LOGICAL-a676-6eaa54533fa6c40716d2e591dd62b845850b3e51886658233aeb356ffba019d03</originalsourceid><addsrcrecordid>eNotz01OwzAUBGBvWKDCAVjhCzjYcey67KrwU6RIRGrENnpJnoOlxEauaentKS2r2YxG8xFyJ3hWGKX4A8Qft89yIfKML0Whr8lY12zzPYPf4vhIy-A99sntXTqy9QEi0jrEFMEleirM6BMkFzw9uPRJgVYQR2TbHiakDU7YB28xou-dH-mHGzDQJ0iww3RDrixMO7z9zwVpXp6bcsOq99e3cl0x0EvNNAKoQklpQffF6aIeclQrMQw670yhjOKdRCWM0VqZXErATiptbQdcrAYuF-T-MnuWtl_RzRCP7Z-4PYvlLyMSUUI</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype></control><display><type>article</type><title>PP-HumanSeg: Connectivity-Aware Portrait Segmentation with a Large-Scale Teleconferencing Video Dataset</title><source>arXiv.org</source><creator>Chu, Lutao ; Liu, Yi ; Wu, Zewu ; Tang, Shiyu ; Chen, Guowei ; Hao, Yuying ; Peng, Juncai ; Yu, Zhiliang ; Chen, Zeyu ; Lai, Baohua ; Xiong, Haoyi</creator><creatorcontrib>Chu, Lutao ; Liu, Yi ; Wu, Zewu ; Tang, Shiyu ; Chen, Guowei ; Hao, Yuying ; Peng, Juncai ; Yu, Zhiliang ; Chen, Zeyu ; Lai, Baohua ; Xiong, Haoyi</creatorcontrib><description>As the COVID-19 pandemic rampages across the world, the demands of video conferencing surge. To this end, real-time portrait segmentation becomes a popular feature to replace backgrounds of conferencing participants. While feature-rich datasets, models and algorithms have been offered for segmentation that extract body postures from life scenes, portrait segmentation has yet not been well covered in a video conferencing context. To facilitate the progress in this field, we introduce an open-source solution named PP-HumanSeg. This work is the first to construct a large-scale video portrait dataset that contains 291 videos from 23 conference scenes with 14K fine-labeled frames and extensions to multi-camera teleconferencing. Furthermore, we propose a novel Semantic Connectivity-aware Learning (SCL) for semantic segmentation, which introduces a semantic connectivity-aware loss to improve the quality of segmentation results from the perspective of connectivity. And we propose an ultra-lightweight model with SCL for practical portrait segmentation, which achieves the best trade-off between IoU and the speed of inference. Extensive evaluations on our dataset demonstrate the superiority of SCL and our model. The source code is available at https://github.com/PaddlePaddle/PaddleSeg.</description><identifier>DOI: 10.48550/arxiv.2112.07146</identifier><language>eng</language><subject>Computer Science - Computer Vision and Pattern Recognition ; Computer Science - Learning</subject><creationdate>2021-12</creationdate><rights>http://arxiv.org/licenses/nonexclusive-distrib/1.0</rights><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>228,230,780,885</link.rule.ids><linktorsrc>$$Uhttps://arxiv.org/abs/2112.07146$$EView_record_in_Cornell_University$$FView_record_in_$$GCornell_University$$Hfree_for_read</linktorsrc><backlink>$$Uhttps://doi.org/10.48550/arXiv.2112.07146$$DView paper in arXiv$$Hfree_for_read</backlink></links><search><creatorcontrib>Chu, Lutao</creatorcontrib><creatorcontrib>Liu, Yi</creatorcontrib><creatorcontrib>Wu, Zewu</creatorcontrib><creatorcontrib>Tang, Shiyu</creatorcontrib><creatorcontrib>Chen, Guowei</creatorcontrib><creatorcontrib>Hao, Yuying</creatorcontrib><creatorcontrib>Peng, Juncai</creatorcontrib><creatorcontrib>Yu, Zhiliang</creatorcontrib><creatorcontrib>Chen, Zeyu</creatorcontrib><creatorcontrib>Lai, Baohua</creatorcontrib><creatorcontrib>Xiong, Haoyi</creatorcontrib><title>PP-HumanSeg: Connectivity-Aware Portrait Segmentation with a Large-Scale Teleconferencing Video Dataset</title><description>As the COVID-19 pandemic rampages across the world, the demands of video conferencing surge. To this end, real-time portrait segmentation becomes a popular feature to replace backgrounds of conferencing participants. While feature-rich datasets, models and algorithms have been offered for segmentation that extract body postures from life scenes, portrait segmentation has yet not been well covered in a video conferencing context. To facilitate the progress in this field, we introduce an open-source solution named PP-HumanSeg. This work is the first to construct a large-scale video portrait dataset that contains 291 videos from 23 conference scenes with 14K fine-labeled frames and extensions to multi-camera teleconferencing. Furthermore, we propose a novel Semantic Connectivity-aware Learning (SCL) for semantic segmentation, which introduces a semantic connectivity-aware loss to improve the quality of segmentation results from the perspective of connectivity. And we propose an ultra-lightweight model with SCL for practical portrait segmentation, which achieves the best trade-off between IoU and the speed of inference. Extensive evaluations on our dataset demonstrate the superiority of SCL and our model. The source code is available at https://github.com/PaddlePaddle/PaddleSeg.</description><subject>Computer Science - Computer Vision and Pattern Recognition</subject><subject>Computer Science - Learning</subject><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2021</creationdate><recordtype>article</recordtype><sourceid>GOX</sourceid><recordid>eNotz01OwzAUBGBvWKDCAVjhCzjYcey67KrwU6RIRGrENnpJnoOlxEauaentKS2r2YxG8xFyJ3hWGKX4A8Qft89yIfKML0Whr8lY12zzPYPf4vhIy-A99sntXTqy9QEi0jrEFMEleirM6BMkFzw9uPRJgVYQR2TbHiakDU7YB28xou-dH-mHGzDQJ0iww3RDrixMO7z9zwVpXp6bcsOq99e3cl0x0EvNNAKoQklpQffF6aIeclQrMQw670yhjOKdRCWM0VqZXErATiptbQdcrAYuF-T-MnuWtl_RzRCP7Z-4PYvlLyMSUUI</recordid><startdate>20211213</startdate><enddate>20211213</enddate><creator>Chu, Lutao</creator><creator>Liu, Yi</creator><creator>Wu, Zewu</creator><creator>Tang, Shiyu</creator><creator>Chen, Guowei</creator><creator>Hao, Yuying</creator><creator>Peng, Juncai</creator><creator>Yu, Zhiliang</creator><creator>Chen, Zeyu</creator><creator>Lai, Baohua</creator><creator>Xiong, Haoyi</creator><scope>AKY</scope><scope>GOX</scope></search><sort><creationdate>20211213</creationdate><title>PP-HumanSeg: Connectivity-Aware Portrait Segmentation with a Large-Scale Teleconferencing Video Dataset</title><author>Chu, Lutao ; Liu, Yi ; Wu, Zewu ; Tang, Shiyu ; Chen, Guowei ; Hao, Yuying ; Peng, Juncai ; Yu, Zhiliang ; Chen, Zeyu ; Lai, Baohua ; Xiong, Haoyi</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-a676-6eaa54533fa6c40716d2e591dd62b845850b3e51886658233aeb356ffba019d03</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2021</creationdate><topic>Computer Science - Computer Vision and Pattern Recognition</topic><topic>Computer Science - Learning</topic><toplevel>online_resources</toplevel><creatorcontrib>Chu, Lutao</creatorcontrib><creatorcontrib>Liu, Yi</creatorcontrib><creatorcontrib>Wu, Zewu</creatorcontrib><creatorcontrib>Tang, Shiyu</creatorcontrib><creatorcontrib>Chen, Guowei</creatorcontrib><creatorcontrib>Hao, Yuying</creatorcontrib><creatorcontrib>Peng, Juncai</creatorcontrib><creatorcontrib>Yu, Zhiliang</creatorcontrib><creatorcontrib>Chen, Zeyu</creatorcontrib><creatorcontrib>Lai, Baohua</creatorcontrib><creatorcontrib>Xiong, Haoyi</creatorcontrib><collection>arXiv Computer Science</collection><collection>arXiv.org</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Chu, Lutao</au><au>Liu, Yi</au><au>Wu, Zewu</au><au>Tang, Shiyu</au><au>Chen, Guowei</au><au>Hao, Yuying</au><au>Peng, Juncai</au><au>Yu, Zhiliang</au><au>Chen, Zeyu</au><au>Lai, Baohua</au><au>Xiong, Haoyi</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>PP-HumanSeg: Connectivity-Aware Portrait Segmentation with a Large-Scale Teleconferencing Video Dataset</atitle><date>2021-12-13</date><risdate>2021</risdate><abstract>As the COVID-19 pandemic rampages across the world, the demands of video conferencing surge. To this end, real-time portrait segmentation becomes a popular feature to replace backgrounds of conferencing participants. While feature-rich datasets, models and algorithms have been offered for segmentation that extract body postures from life scenes, portrait segmentation has yet not been well covered in a video conferencing context. To facilitate the progress in this field, we introduce an open-source solution named PP-HumanSeg. This work is the first to construct a large-scale video portrait dataset that contains 291 videos from 23 conference scenes with 14K fine-labeled frames and extensions to multi-camera teleconferencing. Furthermore, we propose a novel Semantic Connectivity-aware Learning (SCL) for semantic segmentation, which introduces a semantic connectivity-aware loss to improve the quality of segmentation results from the perspective of connectivity. And we propose an ultra-lightweight model with SCL for practical portrait segmentation, which achieves the best trade-off between IoU and the speed of inference. Extensive evaluations on our dataset demonstrate the superiority of SCL and our model. The source code is available at https://github.com/PaddlePaddle/PaddleSeg.</abstract><doi>10.48550/arxiv.2112.07146</doi><oa>free_for_read</oa></addata></record>
fulltext fulltext_linktorsrc
identifier DOI: 10.48550/arxiv.2112.07146
ispartof
issn
language eng
recordid cdi_arxiv_primary_2112_07146
source arXiv.org
subjects Computer Science - Computer Vision and Pattern Recognition
Computer Science - Learning
title PP-HumanSeg: Connectivity-Aware Portrait Segmentation with a Large-Scale Teleconferencing Video Dataset
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-08T11%3A31%3A04IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-arxiv_GOX&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=PP-HumanSeg:%20Connectivity-Aware%20Portrait%20Segmentation%20with%20a%20Large-Scale%20Teleconferencing%20Video%20Dataset&rft.au=Chu,%20Lutao&rft.date=2021-12-13&rft_id=info:doi/10.48550/arxiv.2112.07146&rft_dat=%3Carxiv_GOX%3E2112_07146%3C/arxiv_GOX%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_id=info:pmid/&rfr_iscdi=true