Progressive Volume Distillation with Active Learning for Efficient NeRF Architecture Conversion

Neural Radiance Fields (NeRF) have been widely adopted as practical and versatile representations for 3D scenes, facilitating various downstream tasks. However, different architectures, including the plain Multi-Layer Perceptron (MLP), Tensors, low-rank Tensors, Hashtables, and their combinations, e...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Fang, Shuangkang, Wang, Yufeng, Yang, Yi, Xu, Weixin, Wang, Heng, Ding, Wenrui, Zhou, Shuchang
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 Fang, Shuangkang
Wang, Yufeng
Yang, Yi
Xu, Weixin
Wang, Heng
Ding, Wenrui
Zhou, Shuchang
description Neural Radiance Fields (NeRF) have been widely adopted as practical and versatile representations for 3D scenes, facilitating various downstream tasks. However, different architectures, including the plain Multi-Layer Perceptron (MLP), Tensors, low-rank Tensors, Hashtables, and their combinations, entail distinct trade-offs. For instance, representations based on Hashtables enable faster rendering but lack clear geometric meaning, thereby posing challenges for spatial-relation-aware editing. To address this limitation and maximize the potential of each architecture, we propose Progressive Volume Distillation with Active Learning (PVD-AL), a systematic distillation method that enables any-to-any conversion between diverse architectures. PVD-AL decomposes each structure into two parts and progressively performs distillation from shallower to deeper volume representation, leveraging effective information retrieved from the rendering process. Additionally, a three-level active learning technique provides continuous feedback from teacher to student during the distillation process, achieving high-performance outcomes. Experimental evidence showcases the effectiveness of our method across multiple benchmark datasets. For instance, PVD-AL can distill an MLP-based model from a Hashtables-based model at a 10~20X faster speed and 0.8dB~2dB higher PSNR than training the MLP-based model from scratch. Moreover, PVD-AL permits the fusion of diverse features among distinct structures, enabling models with multiple editing properties and providing a more efficient model to meet real-time requirements like mobile devices. Project website: https://sk-fun.fun/PVD-AL.
doi_str_mv 10.48550/arxiv.2304.04012
format Article
fullrecord <record><control><sourceid>arxiv_GOX</sourceid><recordid>TN_cdi_arxiv_primary_2304_04012</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2304_04012</sourcerecordid><originalsourceid>FETCH-arxiv_primary_2304_040123</originalsourceid><addsrcrecordid>eNqFzr0KwkAQBOBrLER9ACv3BYyXP7ANMcFCRERswxE2yUJyJ3uXqG-vCfZWU8wMfEKsfelF-ziWO8UvGrwglJEnI-kHc1Fc2NSM1tKAcDdt3yEcyDpqW-XIaHiSayAp3difULEmXUNlGLKqopJQOzjjNYeEy4Yclq5nhNToAdl-_0sxq1RrcfXLhdjk2S09bidK8WDqFL-LkVRMpPD_4gNYCUJ0</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype></control><display><type>article</type><title>Progressive Volume Distillation with Active Learning for Efficient NeRF Architecture Conversion</title><source>arXiv.org</source><creator>Fang, Shuangkang ; Wang, Yufeng ; Yang, Yi ; Xu, Weixin ; Wang, Heng ; Ding, Wenrui ; Zhou, Shuchang</creator><creatorcontrib>Fang, Shuangkang ; Wang, Yufeng ; Yang, Yi ; Xu, Weixin ; Wang, Heng ; Ding, Wenrui ; Zhou, Shuchang</creatorcontrib><description>Neural Radiance Fields (NeRF) have been widely adopted as practical and versatile representations for 3D scenes, facilitating various downstream tasks. However, different architectures, including the plain Multi-Layer Perceptron (MLP), Tensors, low-rank Tensors, Hashtables, and their combinations, entail distinct trade-offs. For instance, representations based on Hashtables enable faster rendering but lack clear geometric meaning, thereby posing challenges for spatial-relation-aware editing. To address this limitation and maximize the potential of each architecture, we propose Progressive Volume Distillation with Active Learning (PVD-AL), a systematic distillation method that enables any-to-any conversion between diverse architectures. PVD-AL decomposes each structure into two parts and progressively performs distillation from shallower to deeper volume representation, leveraging effective information retrieved from the rendering process. Additionally, a three-level active learning technique provides continuous feedback from teacher to student during the distillation process, achieving high-performance outcomes. Experimental evidence showcases the effectiveness of our method across multiple benchmark datasets. For instance, PVD-AL can distill an MLP-based model from a Hashtables-based model at a 10~20X faster speed and 0.8dB~2dB higher PSNR than training the MLP-based model from scratch. Moreover, PVD-AL permits the fusion of diverse features among distinct structures, enabling models with multiple editing properties and providing a more efficient model to meet real-time requirements like mobile devices. Project website: https://sk-fun.fun/PVD-AL.</description><identifier>DOI: 10.48550/arxiv.2304.04012</identifier><language>eng</language><subject>Computer Science - Computer Vision and Pattern Recognition</subject><creationdate>2023-04</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/2304.04012$$EView_record_in_Cornell_University$$FView_record_in_$$GCornell_University$$Hfree_for_read</linktorsrc><backlink>$$Uhttps://doi.org/10.48550/arXiv.2304.04012$$DView paper in arXiv$$Hfree_for_read</backlink></links><search><creatorcontrib>Fang, Shuangkang</creatorcontrib><creatorcontrib>Wang, Yufeng</creatorcontrib><creatorcontrib>Yang, Yi</creatorcontrib><creatorcontrib>Xu, Weixin</creatorcontrib><creatorcontrib>Wang, Heng</creatorcontrib><creatorcontrib>Ding, Wenrui</creatorcontrib><creatorcontrib>Zhou, Shuchang</creatorcontrib><title>Progressive Volume Distillation with Active Learning for Efficient NeRF Architecture Conversion</title><description>Neural Radiance Fields (NeRF) have been widely adopted as practical and versatile representations for 3D scenes, facilitating various downstream tasks. However, different architectures, including the plain Multi-Layer Perceptron (MLP), Tensors, low-rank Tensors, Hashtables, and their combinations, entail distinct trade-offs. For instance, representations based on Hashtables enable faster rendering but lack clear geometric meaning, thereby posing challenges for spatial-relation-aware editing. To address this limitation and maximize the potential of each architecture, we propose Progressive Volume Distillation with Active Learning (PVD-AL), a systematic distillation method that enables any-to-any conversion between diverse architectures. PVD-AL decomposes each structure into two parts and progressively performs distillation from shallower to deeper volume representation, leveraging effective information retrieved from the rendering process. Additionally, a three-level active learning technique provides continuous feedback from teacher to student during the distillation process, achieving high-performance outcomes. Experimental evidence showcases the effectiveness of our method across multiple benchmark datasets. For instance, PVD-AL can distill an MLP-based model from a Hashtables-based model at a 10~20X faster speed and 0.8dB~2dB higher PSNR than training the MLP-based model from scratch. Moreover, PVD-AL permits the fusion of diverse features among distinct structures, enabling models with multiple editing properties and providing a more efficient model to meet real-time requirements like mobile devices. Project website: https://sk-fun.fun/PVD-AL.</description><subject>Computer Science - Computer Vision and Pattern Recognition</subject><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2023</creationdate><recordtype>article</recordtype><sourceid>GOX</sourceid><recordid>eNqFzr0KwkAQBOBrLER9ACv3BYyXP7ANMcFCRERswxE2yUJyJ3uXqG-vCfZWU8wMfEKsfelF-ziWO8UvGrwglJEnI-kHc1Fc2NSM1tKAcDdt3yEcyDpqW-XIaHiSayAp3difULEmXUNlGLKqopJQOzjjNYeEy4Yclq5nhNToAdl-_0sxq1RrcfXLhdjk2S09bidK8WDqFL-LkVRMpPD_4gNYCUJ0</recordid><startdate>20230408</startdate><enddate>20230408</enddate><creator>Fang, Shuangkang</creator><creator>Wang, Yufeng</creator><creator>Yang, Yi</creator><creator>Xu, Weixin</creator><creator>Wang, Heng</creator><creator>Ding, Wenrui</creator><creator>Zhou, Shuchang</creator><scope>AKY</scope><scope>GOX</scope></search><sort><creationdate>20230408</creationdate><title>Progressive Volume Distillation with Active Learning for Efficient NeRF Architecture Conversion</title><author>Fang, Shuangkang ; Wang, Yufeng ; Yang, Yi ; Xu, Weixin ; Wang, Heng ; Ding, Wenrui ; Zhou, Shuchang</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-arxiv_primary_2304_040123</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2023</creationdate><topic>Computer Science - Computer Vision and Pattern Recognition</topic><toplevel>online_resources</toplevel><creatorcontrib>Fang, Shuangkang</creatorcontrib><creatorcontrib>Wang, Yufeng</creatorcontrib><creatorcontrib>Yang, Yi</creatorcontrib><creatorcontrib>Xu, Weixin</creatorcontrib><creatorcontrib>Wang, Heng</creatorcontrib><creatorcontrib>Ding, Wenrui</creatorcontrib><creatorcontrib>Zhou, Shuchang</creatorcontrib><collection>arXiv Computer Science</collection><collection>arXiv.org</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Fang, Shuangkang</au><au>Wang, Yufeng</au><au>Yang, Yi</au><au>Xu, Weixin</au><au>Wang, Heng</au><au>Ding, Wenrui</au><au>Zhou, Shuchang</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Progressive Volume Distillation with Active Learning for Efficient NeRF Architecture Conversion</atitle><date>2023-04-08</date><risdate>2023</risdate><abstract>Neural Radiance Fields (NeRF) have been widely adopted as practical and versatile representations for 3D scenes, facilitating various downstream tasks. However, different architectures, including the plain Multi-Layer Perceptron (MLP), Tensors, low-rank Tensors, Hashtables, and their combinations, entail distinct trade-offs. For instance, representations based on Hashtables enable faster rendering but lack clear geometric meaning, thereby posing challenges for spatial-relation-aware editing. To address this limitation and maximize the potential of each architecture, we propose Progressive Volume Distillation with Active Learning (PVD-AL), a systematic distillation method that enables any-to-any conversion between diverse architectures. PVD-AL decomposes each structure into two parts and progressively performs distillation from shallower to deeper volume representation, leveraging effective information retrieved from the rendering process. Additionally, a three-level active learning technique provides continuous feedback from teacher to student during the distillation process, achieving high-performance outcomes. Experimental evidence showcases the effectiveness of our method across multiple benchmark datasets. For instance, PVD-AL can distill an MLP-based model from a Hashtables-based model at a 10~20X faster speed and 0.8dB~2dB higher PSNR than training the MLP-based model from scratch. Moreover, PVD-AL permits the fusion of diverse features among distinct structures, enabling models with multiple editing properties and providing a more efficient model to meet real-time requirements like mobile devices. Project website: https://sk-fun.fun/PVD-AL.</abstract><doi>10.48550/arxiv.2304.04012</doi><oa>free_for_read</oa></addata></record>
fulltext fulltext_linktorsrc
identifier DOI: 10.48550/arxiv.2304.04012
ispartof
issn
language eng
recordid cdi_arxiv_primary_2304_04012
source arXiv.org
subjects Computer Science - Computer Vision and Pattern Recognition
title Progressive Volume Distillation with Active Learning for Efficient NeRF Architecture Conversion
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-12T02%3A01%3A28IST&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=Progressive%20Volume%20Distillation%20with%20Active%20Learning%20for%20Efficient%20NeRF%20Architecture%20Conversion&rft.au=Fang,%20Shuangkang&rft.date=2023-04-08&rft_id=info:doi/10.48550/arxiv.2304.04012&rft_dat=%3Carxiv_GOX%3E2304_04012%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