{"id":"https://openalex.org/W7127623234","doi":"https://doi.org/10.48550/arxiv.2602.03615","title":"KTV: Keyframes and Key Tokens Selection for Efficient Training-Free Video LLMs","display_name":"KTV: Keyframes and Key Tokens Selection for Efficient Training-Free Video LLMs","publication_year":2026,"publication_date":"2026-02-03","ids":{"openalex":"https://openalex.org/W7127623234","doi":"https://doi.org/10.48550/arxiv.2602.03615"},"language":null,"primary_location":{"id":"pmh:doi:10.48550/arxiv.2602.03615","is_oa":true,"landing_page_url":null,"pdf_url":null,"source":{"id":"https://openalex.org/S4406922384","display_name":"Open MIND","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":"publisher-specific-oa","license_id":"https://openalex.org/licenses/publisher-specific-oa","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Article"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":null,"any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5125070953","display_name":"Baiyang Song","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Song, Baiyang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5125095941","display_name":"Jun Peng","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Peng, Jun","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5125050904","display_name":"Yuxin Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Yuxin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5125029289","display_name":"Guangyao Chen","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chen, Guangyao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5040168310","display_name":"Feidiao Yang","orcid":"https://orcid.org/0000-0001-7637-2562"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yang, Feidiao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5124998167","display_name":"Jianyuan Guo","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Guo, Jianyuan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":6,"corresponding_author_ids":[],"corresponding_institution_ids":[],"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":null,"last_page":null},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11714","display_name":"Multimodal Machine Learning Applications","score":0.9581000208854675,"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":0.9581000208854675,"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/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.013000000268220901,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"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/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.008899999782443047,"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":0.6191999912261963},{"id":"https://openalex.org/keywords/selection","display_name":"Selection (genetic algorithm)","score":0.6079999804496765},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.593999981880188},{"id":"https://openalex.org/keywords/redundancy","display_name":"Redundancy (engineering)","score":0.5644999742507935},{"id":"https://openalex.org/keywords/task","display_name":"Task (project management)","score":0.5496000051498413},{"id":"https://openalex.org/keywords/pruning","display_name":"Pruning","score":0.5274999737739563},{"id":"https://openalex.org/keywords/rapid-serial-visual-presentation","display_name":"Rapid serial visual presentation","score":0.4562000036239624},{"id":"https://openalex.org/keywords/cluster-analysis","display_name":"Cluster analysis","score":0.4507000148296356},{"id":"https://openalex.org/keywords/visualization","display_name":"Visualization","score":0.3695000112056732}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8331000208854675},{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.6191999912261963},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.6158000230789185},{"id":"https://openalex.org/C81917197","wikidata":"https://www.wikidata.org/wiki/Q628760","display_name":"Selection (genetic algorithm)","level":2,"score":0.6079999804496765},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.593999981880188},{"id":"https://openalex.org/C152124472","wikidata":"https://www.wikidata.org/wiki/Q1204361","display_name":"Redundancy (engineering)","level":2,"score":0.5644999742507935},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.5496000051498413},{"id":"https://openalex.org/C108010975","wikidata":"https://www.wikidata.org/wiki/Q500094","display_name":"Pruning","level":2,"score":0.5274999737739563},{"id":"https://openalex.org/C2779668609","wikidata":"https://www.wikidata.org/wiki/Q623092","display_name":"Rapid serial visual presentation","level":3,"score":0.4562000036239624},{"id":"https://openalex.org/C73555534","wikidata":"https://www.wikidata.org/wiki/Q622825","display_name":"Cluster analysis","level":2,"score":0.4507000148296356},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.41019999980926514},{"id":"https://openalex.org/C36464697","wikidata":"https://www.wikidata.org/wiki/Q451553","display_name":"Visualization","level":2,"score":0.3695000112056732},{"id":"https://openalex.org/C202474056","wikidata":"https://www.wikidata.org/wiki/Q1931635","display_name":"Video tracking","level":3,"score":0.3675999939441681},{"id":"https://openalex.org/C31972630","wikidata":"https://www.wikidata.org/wiki/Q844240","display_name":"Computer vision","level":1,"score":0.35019999742507935},{"id":"https://openalex.org/C2778112365","wikidata":"https://www.wikidata.org/wiki/Q3511065","display_name":"Sequence (biology)","level":2,"score":0.3357999920845032},{"id":"https://openalex.org/C2780929884","wikidata":"https://www.wikidata.org/wiki/Q737460","display_name":"Blindness","level":2,"score":0.3301999866962433},{"id":"https://openalex.org/C158495155","wikidata":"https://www.wikidata.org/wiki/Q2369151","display_name":"Visual search","level":2,"score":0.32269999384880066},{"id":"https://openalex.org/C175154964","wikidata":"https://www.wikidata.org/wiki/Q380077","display_name":"Task analysis","level":3,"score":0.3199000060558319},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.3003000020980835},{"id":"https://openalex.org/C65483669","wikidata":"https://www.wikidata.org/wiki/Q3536669","display_name":"Video processing","level":2,"score":0.2996000051498413},{"id":"https://openalex.org/C124304363","wikidata":"https://www.wikidata.org/wiki/Q673661","display_name":"Abstraction","level":2,"score":0.2870999872684479},{"id":"https://openalex.org/C179799912","wikidata":"https://www.wikidata.org/wiki/Q205084","display_name":"Computational complexity theory","level":2,"score":0.2745000123977661},{"id":"https://openalex.org/C148483581","wikidata":"https://www.wikidata.org/wiki/Q446488","display_name":"Feature selection","level":2,"score":0.2732999920845032},{"id":"https://openalex.org/C153083717","wikidata":"https://www.wikidata.org/wiki/Q6535263","display_name":"Leverage (statistics)","level":2,"score":0.257099986076355},{"id":"https://openalex.org/C178253425","wikidata":"https://www.wikidata.org/wiki/Q162668","display_name":"Visual perception","level":3,"score":0.25529998540878296},{"id":"https://openalex.org/C113954288","wikidata":"https://www.wikidata.org/wiki/Q186885","display_name":"Timestamp","level":2,"score":0.250900000333786}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:doi:10.48550/arxiv.2602.03615","is_oa":true,"landing_page_url":null,"pdf_url":null,"source":{"id":"https://openalex.org/S4406922384","display_name":"Open MIND","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":"publisher-specific-oa","license_id":"https://openalex.org/licenses/publisher-specific-oa","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Article"},{"id":"doi:10.48550/arxiv.2602.03615","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2602.03615","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":"article"}],"best_oa_location":{"id":"pmh:doi:10.48550/arxiv.2602.03615","is_oa":true,"landing_page_url":null,"pdf_url":null,"source":{"id":"https://openalex.org/S4406922384","display_name":"Open MIND","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":"publisher-specific-oa","license_id":"https://openalex.org/licenses/publisher-specific-oa","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Article"},"sustainable_development_goals":[{"id":"https://metadata.un.org/sdg/4","display_name":"Quality Education","score":0.4174593985080719}],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Training-free":[0],"video":[1,18,75,94,188],"understanding":[2],"leverages":[3],"the":[4,27,97,124,152,158,163,196],"strong":[5],"image":[6],"comprehension":[7],"capabilities":[8],"of":[9,22,117,154],"pre-trained":[10],"vision":[11],"language":[12],"models":[13],"(VLMs)":[14],"by":[15,105],"treating":[16],"a":[17,20,85,111,186],"as":[19],"sequence":[21],"static":[23],"frames,":[24,71,191],"thus":[25],"obviating":[26],"need":[28],"for":[29,89,185],"costly":[30],"video-specific":[31],"training.":[32],"However,":[33],"this":[34],"paradigm":[35],"often":[36],"suffers":[37],"from":[38,139],"severe":[39],"visual":[40,108,130,178,183],"redundancy":[41],"and":[42,66,91,114,147],"high":[43],"computational":[44],"overhead,":[45],"especially":[46,56],"when":[47],"processing":[48],"long":[49],"videos.":[50],"Crucially,":[51],"existing":[52],"keyframe":[53,103,142],"selection":[54,104],"strategies,":[55],"those":[57],"based":[58,143],"on":[59,144,162,195,207],"CLIP":[60],"similarity,":[61],"are":[62],"prone":[63],"to":[64],"biases":[65],"may":[67],"inadvertently":[68],"overlook":[69],"critical":[70],"resulting":[72],"in":[73],"suboptimal":[74],"comprehension.":[76],"To":[77],"address":[78],"these":[79],"significant":[80],"challenges,":[81],"we":[82],"propose":[83],"\\textbf{KTV},":[84],"novel":[86],"two-stage":[87],"framework":[88],"efficient":[90],"effective":[92],"training-free":[93,172],"understanding.":[95],"In":[96,123,199],"first":[98],"stage,":[99,126],"KTV":[100,127,169,201],"performs":[101],"question-agnostic":[102],"clustering":[106],"frame-level":[107],"features,":[109],"yielding":[110],"compact,":[112],"diverse,":[113],"representative":[115],"subset":[116],"frames":[118],"that":[119,168],"mitigates":[120],"temporal":[121],"redundancy.":[122],"second":[125],"applies":[128],"key":[129],"token":[131,145],"selection,":[132],"pruning":[133],"redundant":[134],"or":[135],"less":[136],"informative":[137],"tokens":[138,155,184],"each":[140],"selected":[141],"importance":[146],"redundancy,":[148],"which":[149],"significantly":[150,176],"reduces":[151],"number":[153],"fed":[156],"into":[157],"LLM.":[159],"Extensive":[160],"experiments":[161],"Multiple-Choice":[164],"VideoQA":[165],"task":[166],"demonstrate":[167],"outperforms":[170],"state-of-the-art":[171],"baselines":[173],"while":[174],"using":[175],"fewer":[177],"tokens,":[179],"\\emph{e.g.},":[180],"only":[181],"504":[182],"60-min":[187],"with":[189],"10800":[190],"achieving":[192],"$44.8\\%$":[193],"accuracy":[194],"MLVU-Test":[197],"benchmark.":[198],"particular,":[200],"also":[202],"exceeds":[203],"several":[204],"training-based":[205],"approaches":[206],"certain":[208],"benchmarks.":[209]},"counts_by_year":[],"updated_date":"2026-06-11T09:08:48.828518","created_date":"2026-02-06T00:00:00"}