{"id":"https://openalex.org/W4417280611","doi":"https://doi.org/10.1109/iccv51701.2025.01934","title":"Keyframe-Oriented Vision Token Pruning: Enhancing Efficiency of Large Vision Language Models on Long-form Video Processing","display_name":"Keyframe-Oriented Vision Token Pruning: Enhancing Efficiency of Large Vision Language Models on Long-form Video Processing","publication_year":2025,"publication_date":"2025-10-19","ids":{"openalex":"https://openalex.org/W4417280611","doi":"https://doi.org/10.1109/iccv51701.2025.01934"},"language":"en","primary_location":{"id":"doi:10.1109/iccv51701.2025.01934","is_oa":false,"landing_page_url":"https://doi.org/10.1109/iccv51701.2025.01934","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2025 IEEE/CVF International Conference on Computer Vision (ICCV)","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["arxiv","crossref","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2503.10742","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5079115940","display_name":"Yudong Liu","orcid":"https://orcid.org/0000-0002-5153-3400"},"institutions":[{"id":"https://openalex.org/I170897317","display_name":"Duke University","ror":"https://ror.org/00py81415","country_code":"US","type":"education","lineage":["https://openalex.org/I170897317"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Yudong Liu","raw_affiliation_strings":["Duke University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Duke University","institution_ids":["https://openalex.org/I170897317"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5024963902","display_name":"Jingwei Sun","orcid":"https://orcid.org/0000-0001-8358-9487"},"institutions":[{"id":"https://openalex.org/I170897317","display_name":"Duke University","ror":"https://ror.org/00py81415","country_code":"US","type":"education","lineage":["https://openalex.org/I170897317"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Jingwei Sun","raw_affiliation_strings":["Duke University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Duke University","institution_ids":["https://openalex.org/I170897317"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100768194","display_name":"Yueqian Lin","orcid":"https://orcid.org/0000-0003-1473-8981"},"institutions":[{"id":"https://openalex.org/I170897317","display_name":"Duke University","ror":"https://ror.org/00py81415","country_code":"US","type":"education","lineage":["https://openalex.org/I170897317"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Yueqian Lin","raw_affiliation_strings":["Duke University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Duke University","institution_ids":["https://openalex.org/I170897317"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5063092287","display_name":"Jingyang Zhang","orcid":"https://orcid.org/0000-0002-1788-6501"},"institutions":[{"id":"https://openalex.org/I170897317","display_name":"Duke University","ror":"https://ror.org/00py81415","country_code":"US","type":"education","lineage":["https://openalex.org/I170897317"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Jianyi Zhang","raw_affiliation_strings":["Duke University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Duke University","institution_ids":["https://openalex.org/I170897317"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Jingyang Zhang","orcid":null},"institutions":[{"id":"https://openalex.org/I170897317","display_name":"Duke University","ror":"https://ror.org/00py81415","country_code":"US","type":"education","lineage":["https://openalex.org/I170897317"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Jingyang Zhang","raw_affiliation_strings":["Duke University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Duke University","institution_ids":["https://openalex.org/I170897317"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5071294124","display_name":"Ming Yin","orcid":"https://orcid.org/0000-0002-7364-139X"},"institutions":[{"id":"https://openalex.org/I170897317","display_name":"Duke University","ror":"https://ror.org/00py81415","country_code":"US","type":"education","lineage":["https://openalex.org/I170897317"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Ming Yin","raw_affiliation_strings":["Duke University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Duke University","institution_ids":["https://openalex.org/I170897317"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Qinsi Wang","orcid":null},"institutions":[{"id":"https://openalex.org/I170897317","display_name":"Duke University","ror":"https://ror.org/00py81415","country_code":"US","type":"education","lineage":["https://openalex.org/I170897317"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Qinsi Wang","raw_affiliation_strings":["Duke University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Duke University","institution_ids":["https://openalex.org/I170897317"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100429403","display_name":"Hai Li","orcid":"https://orcid.org/0000-0003-3228-6544"},"institutions":[{"id":"https://openalex.org/I170897317","display_name":"Duke University","ror":"https://ror.org/00py81415","country_code":"US","type":"education","lineage":["https://openalex.org/I170897317"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Hai Li","raw_affiliation_strings":["Duke University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Duke University","institution_ids":["https://openalex.org/I170897317"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5102868511","display_name":"Yiran Chen","orcid":"https://orcid.org/0009-0004-4668-4505"},"institutions":[{"id":"https://openalex.org/I170897317","display_name":"Duke University","ror":"https://ror.org/00py81415","country_code":"US","type":"education","lineage":["https://openalex.org/I170897317"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Yiran Chen","raw_affiliation_strings":["Duke University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Duke University","institution_ids":["https://openalex.org/I170897317"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I170897317"],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"20802","last_page":"20811"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11714","display_name":"Multimodal Machine Learning Applications","score":0.9768000245094299,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T11714","display_name":"Multimodal Machine Learning Applications","score":0.9768000245094299,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.004100000020116568,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10627","display_name":"Advanced Image and Video Retrieval Techniques","score":0.0026000000070780516,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/security-token","display_name":"Security token","score":0.7929999828338623},{"id":"https://openalex.org/keywords/pruning","display_name":"Pruning","score":0.6446999907493591},{"id":"https://openalex.org/keywords/scalability","display_name":"Scalability","score":0.5688999891281128},{"id":"https://openalex.org/keywords/overhead","display_name":"Overhead (engineering)","score":0.5619000196456909},{"id":"https://openalex.org/keywords/benchmark","display_name":"Benchmark (surveying)","score":0.49720001220703125},{"id":"https://openalex.org/keywords/focus","display_name":"Focus (optics)","score":0.4936999976634979},{"id":"https://openalex.org/keywords/relevance","display_name":"Relevance (law)","score":0.48179998993873596},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.4449000060558319}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8090000152587891},{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.7929999828338623},{"id":"https://openalex.org/C108010975","wikidata":"https://www.wikidata.org/wiki/Q500094","display_name":"Pruning","level":2,"score":0.6446999907493591},{"id":"https://openalex.org/C48044578","wikidata":"https://www.wikidata.org/wiki/Q727490","display_name":"Scalability","level":2,"score":0.5688999891281128},{"id":"https://openalex.org/C2779960059","wikidata":"https://www.wikidata.org/wiki/Q7113681","display_name":"Overhead (engineering)","level":2,"score":0.5619000196456909},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5529999732971191},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.49720001220703125},{"id":"https://openalex.org/C192209626","wikidata":"https://www.wikidata.org/wiki/Q190909","display_name":"Focus (optics)","level":2,"score":0.4936999976634979},{"id":"https://openalex.org/C158154518","wikidata":"https://www.wikidata.org/wiki/Q7310970","display_name":"Relevance (law)","level":2,"score":0.48179998993873596},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.4449000060558319},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.41670000553131104},{"id":"https://openalex.org/C126042441","wikidata":"https://www.wikidata.org/wiki/Q1324888","display_name":"Frame (networking)","level":2,"score":0.385699987411499},{"id":"https://openalex.org/C31972630","wikidata":"https://www.wikidata.org/wiki/Q844240","display_name":"Computer vision","level":1,"score":0.38429999351501465},{"id":"https://openalex.org/C5339829","wikidata":"https://www.wikidata.org/wiki/Q1425977","display_name":"Machine vision","level":2,"score":0.3319999873638153},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.3221000134944916},{"id":"https://openalex.org/C107457646","wikidata":"https://www.wikidata.org/wiki/Q207434","display_name":"Human\u2013computer interaction","level":1,"score":0.2913999855518341},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.2840000092983246},{"id":"https://openalex.org/C125411270","wikidata":"https://www.wikidata.org/wiki/Q18653","display_name":"Encoding (memory)","level":2,"score":0.28040000796318054},{"id":"https://openalex.org/C66024118","wikidata":"https://www.wikidata.org/wiki/Q1122506","display_name":"Computational model","level":2,"score":0.2671999931335449},{"id":"https://openalex.org/C108583219","wikidata":"https://www.wikidata.org/wiki/Q197536","display_name":"Deep learning","level":2,"score":0.26570001244544983},{"id":"https://openalex.org/C111335779","wikidata":"https://www.wikidata.org/wiki/Q3454686","display_name":"Reduction (mathematics)","level":2,"score":0.26089999079704285},{"id":"https://openalex.org/C36464697","wikidata":"https://www.wikidata.org/wiki/Q451553","display_name":"Visualization","level":2,"score":0.25459998846054077}],"mesh":[],"locations_count":3,"locations":[{"id":"doi:10.1109/iccv51701.2025.01934","is_oa":false,"landing_page_url":"https://doi.org/10.1109/iccv51701.2025.01934","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2025 IEEE/CVF International Conference on Computer Vision (ICCV)","raw_type":"proceedings-article"},{"id":"pmh:oai:arXiv.org:2503.10742","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2503.10742","pdf_url":"https://arxiv.org/pdf/2503.10742","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2503.10742","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2503.10742","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2503.10742","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2503.10742","pdf_url":"https://arxiv.org/pdf/2503.10742","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G6624552157","display_name":null,"funder_award_id":"W911NF-23-2-0224","funder_id":"https://openalex.org/F4320338281","funder_display_name":"Army Research Office"},{"id":"https://openalex.org/G8377316191","display_name":"AI Institute for Edge Computing Leveraging Next Generation Networks (Athena)","funder_award_id":"2112562","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"}],"funders":[{"id":"https://openalex.org/F4320306076","display_name":"National Science Foundation","ror":"https://ror.org/021nxhr62"},{"id":"https://openalex.org/F4320338281","display_name":"Army Research Office","ror":"https://ror.org/05epdh915"}],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Vision":[0,66],"language":[1],"models":[2],"(VLMs)":[3],"demonstrate":[4,171],"strong":[5],"capabilities":[6],"in":[7,27,175],"jointly":[8],"processing":[9],"visual":[10,24],"and":[11,79,117,123,159],"textual":[12],"data.":[13],"However,":[14],"they":[15],"often":[16],"incur":[17],"substantial":[18],"computational":[19],"overhead":[20],"due":[21],"to":[22,91],"redundant":[23,103],"information,":[25],"particularly":[26],"long-form":[28,109],"video":[29,110],"scenarios.":[30],"Existing":[31],"approaches":[32],"predominantly":[33],"focus":[34],"on":[35,88],"either":[36],"vision":[37],"token":[38,77,152],"pruning,":[39],"which":[40,48],"may":[41],"overlook":[42],"spatio-temporal":[43],"dependencies,":[44],"or":[45],"keyframe":[46,80],"selection,":[47],"identifies":[49],"informative":[50],"frames":[51],"but":[52,137],"discards":[53],"others,":[54],"thus":[55],"disrupting":[56],"contextual":[57,98,160],"continuity.":[58],"In":[59],"this":[60],"work,":[61],"we":[62,115],"propose":[63],"KVTP":[64,94,149],"(Keyframe-oriented":[65],"Token":[67],"Pruning),":[68],"a":[69,126],"novel":[70],"framework":[71],"that":[72,131,148],"overcomes":[73],"the":[74,92,108,167],"drawbacks":[75],"of":[76,113,144],"pruning":[78,85],"selection.":[81],"By":[82],"adaptively":[83],"assigning":[84],"rates":[86],"based":[87],"frame":[89],"relevance":[90],"query,":[93],"effectively":[95],"retains":[96],"essential":[97],"information":[99],"while":[100,165],"significantly":[101,162],"reducing":[102],"computation.":[104],"To":[105],"thoroughly":[106],"evaluate":[107],"understanding":[111],"capacities":[112],"VLMs,":[114],"curated":[116],"reorganized":[118],"subsets":[119],"from":[120],"VideoMME,":[121],"EgoSchema,":[122],"NextQA":[124],"into":[125],"unified":[127],"benchmark":[128],"named":[129],"SparseKV-QA":[130],"highlights":[132],"real-world":[133],"scenarios":[134],"with":[135,142],"sparse":[136],"crucial":[138],"events.":[139],"Our":[140],"experiments":[141],"VLMs":[143],"various":[145],"scales":[146],"show":[147],"can":[150],"reduce":[151],"usage":[153],"by":[154],"80%":[155],"without":[156],"compromising":[157],"spatiotemporal":[158],"consistency,":[161],"cutting":[163],"computation":[164],"maintaining":[166],"performance.":[168],"These":[169],"results":[170],"our":[172],"approach's":[173],"effectiveness":[174],"efficient":[176],"long-video":[177],"processing,":[178],"facilitating":[179],"more":[180],"scalable":[181],"VLM":[182],"deployment.":[183]},"counts_by_year":[],"updated_date":"2026-07-21T08:15:58.654021","created_date":"2025-10-10T00:00:00"}