CL-HOI: Cross-Level Human-Object Interaction Distillation from Vision Large Language Models

Human-object interaction (HOI) detection has seen advancements with Vision Language Models (VLMs), but these methods often depend on extensive manual annotations. Vision Large Language Models (VLLMs) can inherently recognize and reason about interactions at the image level but are computationally he...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:arXiv.org 2024-10
Hauptverfasser: Gao, Jianjun, Chen, Cai, Wang, Ruoyu, Liu, Wenyang, Kim-Hui, Yap, Garg, Kratika, Boon-Siew, Han
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 Gao, Jianjun
Chen, Cai
Wang, Ruoyu
Liu, Wenyang
Kim-Hui, Yap
Garg, Kratika
Boon-Siew, Han
description Human-object interaction (HOI) detection has seen advancements with Vision Language Models (VLMs), but these methods often depend on extensive manual annotations. Vision Large Language Models (VLLMs) can inherently recognize and reason about interactions at the image level but are computationally heavy and not designed for instance-level HOI detection. To overcome these limitations, we propose a Cross-Level HOI distillation (CL-HOI) framework, which distills instance-level HOIs from VLLMs image-level understanding without the need for manual annotations. Our approach involves two stages: context distillation, where a Visual Linguistic Translator (VLT) converts visual information into linguistic form, and interaction distillation, where an Interaction Cognition Network (ICN) reasons about spatial, visual, and context relations. We design contrastive distillation losses to transfer image-level context and interaction knowledge from the teacher to the student model, enabling instance-level HOI detection. Evaluations on HICO-DET and V-COCO datasets demonstrate that our CL-HOI surpasses existing weakly supervised methods and VLLM supervised methods, showing its efficacy in detecting HOIs without manual labels.
format Article
fullrecord <record><control><sourceid>proquest</sourceid><recordid>TN_cdi_proquest_journals_3119288673</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>3119288673</sourcerecordid><originalsourceid>FETCH-proquest_journals_31192886733</originalsourceid><addsrcrecordid>eNqNit8KgjAchUcQJOU7DLoe6JZ_6tYKBcOb6KYLWTZlMrfab_b8WfQA3ZzzfZwzQx5lLCTphtIF8gH6IAhonNAoYh66ZiXJq2KHM2sASCleQuF8HLgm1a0XjcOFdsLyxkmj8V6Ck0rxr7TWDPgi4cMlt52YUncjn-Bk7kLBCs1brkD4v16i9fFwznLysOY5CnB1b0arp6lmYbilaRonjP33egOxCEHx</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>3119288673</pqid></control><display><type>article</type><title>CL-HOI: Cross-Level Human-Object Interaction Distillation from Vision Large Language Models</title><source>Free E- Journals</source><creator>Gao, Jianjun ; Chen, Cai ; Wang, Ruoyu ; Liu, Wenyang ; Kim-Hui, Yap ; Garg, Kratika ; Boon-Siew, Han</creator><creatorcontrib>Gao, Jianjun ; Chen, Cai ; Wang, Ruoyu ; Liu, Wenyang ; Kim-Hui, Yap ; Garg, Kratika ; Boon-Siew, Han</creatorcontrib><description>Human-object interaction (HOI) detection has seen advancements with Vision Language Models (VLMs), but these methods often depend on extensive manual annotations. Vision Large Language Models (VLLMs) can inherently recognize and reason about interactions at the image level but are computationally heavy and not designed for instance-level HOI detection. To overcome these limitations, we propose a Cross-Level HOI distillation (CL-HOI) framework, which distills instance-level HOIs from VLLMs image-level understanding without the need for manual annotations. Our approach involves two stages: context distillation, where a Visual Linguistic Translator (VLT) converts visual information into linguistic form, and interaction distillation, where an Interaction Cognition Network (ICN) reasons about spatial, visual, and context relations. We design contrastive distillation losses to transfer image-level context and interaction knowledge from the teacher to the student model, enabling instance-level HOI detection. Evaluations on HICO-DET and V-COCO datasets demonstrate that our CL-HOI surpasses existing weakly supervised methods and VLLM supervised methods, showing its efficacy in detecting HOIs without manual labels.</description><identifier>EISSN: 2331-8422</identifier><language>eng</language><publisher>Ithaca: Cornell University Library, arXiv.org</publisher><subject>Annotations ; Cognition ; Context ; Knowledge management ; Large language models ; Linguistics ; Vision</subject><ispartof>arXiv.org, 2024-10</ispartof><rights>2024. This work is published under http://creativecommons.org/licenses/by/4.0/ (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>Gao, Jianjun</creatorcontrib><creatorcontrib>Chen, Cai</creatorcontrib><creatorcontrib>Wang, Ruoyu</creatorcontrib><creatorcontrib>Liu, Wenyang</creatorcontrib><creatorcontrib>Kim-Hui, Yap</creatorcontrib><creatorcontrib>Garg, Kratika</creatorcontrib><creatorcontrib>Boon-Siew, Han</creatorcontrib><title>CL-HOI: Cross-Level Human-Object Interaction Distillation from Vision Large Language Models</title><title>arXiv.org</title><description>Human-object interaction (HOI) detection has seen advancements with Vision Language Models (VLMs), but these methods often depend on extensive manual annotations. Vision Large Language Models (VLLMs) can inherently recognize and reason about interactions at the image level but are computationally heavy and not designed for instance-level HOI detection. To overcome these limitations, we propose a Cross-Level HOI distillation (CL-HOI) framework, which distills instance-level HOIs from VLLMs image-level understanding without the need for manual annotations. Our approach involves two stages: context distillation, where a Visual Linguistic Translator (VLT) converts visual information into linguistic form, and interaction distillation, where an Interaction Cognition Network (ICN) reasons about spatial, visual, and context relations. We design contrastive distillation losses to transfer image-level context and interaction knowledge from the teacher to the student model, enabling instance-level HOI detection. Evaluations on HICO-DET and V-COCO datasets demonstrate that our CL-HOI surpasses existing weakly supervised methods and VLLM supervised methods, showing its efficacy in detecting HOIs without manual labels.</description><subject>Annotations</subject><subject>Cognition</subject><subject>Context</subject><subject>Knowledge management</subject><subject>Large language models</subject><subject>Linguistics</subject><subject>Vision</subject><issn>2331-8422</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2024</creationdate><recordtype>article</recordtype><sourceid>ABUWG</sourceid><sourceid>AFKRA</sourceid><sourceid>AZQEC</sourceid><sourceid>BENPR</sourceid><sourceid>CCPQU</sourceid><sourceid>DWQXO</sourceid><recordid>eNqNit8KgjAchUcQJOU7DLoe6JZ_6tYKBcOb6KYLWTZlMrfab_b8WfQA3ZzzfZwzQx5lLCTphtIF8gH6IAhonNAoYh66ZiXJq2KHM2sASCleQuF8HLgm1a0XjcOFdsLyxkmj8V6Ck0rxr7TWDPgi4cMlt52YUncjn-Bk7kLBCs1brkD4v16i9fFwznLysOY5CnB1b0arp6lmYbilaRonjP33egOxCEHx</recordid><startdate>20241021</startdate><enddate>20241021</enddate><creator>Gao, Jianjun</creator><creator>Chen, Cai</creator><creator>Wang, Ruoyu</creator><creator>Liu, Wenyang</creator><creator>Kim-Hui, Yap</creator><creator>Garg, Kratika</creator><creator>Boon-Siew, Han</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>20241021</creationdate><title>CL-HOI: Cross-Level Human-Object Interaction Distillation from Vision Large Language Models</title><author>Gao, Jianjun ; Chen, Cai ; Wang, Ruoyu ; Liu, Wenyang ; Kim-Hui, Yap ; Garg, Kratika ; Boon-Siew, Han</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-proquest_journals_31192886733</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2024</creationdate><topic>Annotations</topic><topic>Cognition</topic><topic>Context</topic><topic>Knowledge management</topic><topic>Large language models</topic><topic>Linguistics</topic><topic>Vision</topic><toplevel>online_resources</toplevel><creatorcontrib>Gao, Jianjun</creatorcontrib><creatorcontrib>Chen, Cai</creatorcontrib><creatorcontrib>Wang, Ruoyu</creatorcontrib><creatorcontrib>Liu, Wenyang</creatorcontrib><creatorcontrib>Kim-Hui, Yap</creatorcontrib><creatorcontrib>Garg, Kratika</creatorcontrib><creatorcontrib>Boon-Siew, Han</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 (ProQuest)</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>Gao, Jianjun</au><au>Chen, Cai</au><au>Wang, Ruoyu</au><au>Liu, Wenyang</au><au>Kim-Hui, Yap</au><au>Garg, Kratika</au><au>Boon-Siew, Han</au><format>book</format><genre>document</genre><ristype>GEN</ristype><atitle>CL-HOI: Cross-Level Human-Object Interaction Distillation from Vision Large Language Models</atitle><jtitle>arXiv.org</jtitle><date>2024-10-21</date><risdate>2024</risdate><eissn>2331-8422</eissn><abstract>Human-object interaction (HOI) detection has seen advancements with Vision Language Models (VLMs), but these methods often depend on extensive manual annotations. Vision Large Language Models (VLLMs) can inherently recognize and reason about interactions at the image level but are computationally heavy and not designed for instance-level HOI detection. To overcome these limitations, we propose a Cross-Level HOI distillation (CL-HOI) framework, which distills instance-level HOIs from VLLMs image-level understanding without the need for manual annotations. Our approach involves two stages: context distillation, where a Visual Linguistic Translator (VLT) converts visual information into linguistic form, and interaction distillation, where an Interaction Cognition Network (ICN) reasons about spatial, visual, and context relations. We design contrastive distillation losses to transfer image-level context and interaction knowledge from the teacher to the student model, enabling instance-level HOI detection. Evaluations on HICO-DET and V-COCO datasets demonstrate that our CL-HOI surpasses existing weakly supervised methods and VLLM supervised methods, showing its efficacy in detecting HOIs without manual labels.</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, 2024-10
issn 2331-8422
language eng
recordid cdi_proquest_journals_3119288673
source Free E- Journals
subjects Annotations
Cognition
Context
Knowledge management
Large language models
Linguistics
Vision
title CL-HOI: Cross-Level Human-Object Interaction Distillation from Vision Large Language Models
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-09T03%3A39%3A04IST&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=CL-HOI:%20Cross-Level%20Human-Object%20Interaction%20Distillation%20from%20Vision%20Large%20Language%20Models&rft.jtitle=arXiv.org&rft.au=Gao,%20Jianjun&rft.date=2024-10-21&rft.eissn=2331-8422&rft_id=info:doi/&rft_dat=%3Cproquest%3E3119288673%3C/proquest%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_pqid=3119288673&rft_id=info:pmid/&rfr_iscdi=true