{"id":"https://openalex.org/W7214939883","doi":"https://doi.org/10.48550/arxiv.2609.37950","title":"Video-RSI: Recursive Self-Improvement of Video Understanding Agents via Harness Evolution","display_name":"Video-RSI: Recursive Self-Improvement of Video Understanding Agents via Harness Evolution","publication_year":2026,"publication_date":"2026-09-29","ids":{"openalex":"https://openalex.org/W7214939883","doi":"https://doi.org/10.48550/arxiv.2609.37950"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2609.37950","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2609.37950","pdf_url":"https://arxiv.org/pdf/2609.37950","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"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":"other-oa","license_id":"https://openalex.org/licenses/other-oa","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"volume":null,"type":"preprint","indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2609.37950","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5136762382","display_name":"Bingjun Luo","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I99065089","display_name":"Tsinghua University","ror":"https://ror.org/03cve4549","country_code":"CN","type":"education","lineage":["https://openalex.org/I99065089"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Luo, Bingjun","raw_affiliation_strings":["Tsinghua University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Tsinghua University","institution_ids":["https://openalex.org/I99065089"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101955305","display_name":"Jialin Guo","orcid":"https://orcid.org/0000-0002-7385-6374","observed_orcids":["https://orcid.org/0000-0002-7385-6374"]},"institutions":[{"id":"https://openalex.org/I151727225","display_name":"Harbin Engineering University","ror":"https://ror.org/03x80pn82","country_code":"CN","type":"education","lineage":["https://openalex.org/I151727225"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Guo, Jialin","raw_affiliation_strings":["Harbin Engineering University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Harbin Engineering University","institution_ids":["https://openalex.org/I151727225"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5153672974","display_name":"Siqi Li","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I99065089","display_name":"Tsinghua University","ror":"https://ror.org/03cve4549","country_code":"CN","type":"education","lineage":["https://openalex.org/I99065089"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Li, Siqi","raw_affiliation_strings":["Tsinghua University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Tsinghua University","institution_ids":["https://openalex.org/I99065089"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":2,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"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":0.7958999872207642,"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":0.7958999872207642,"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/T10812","display_name":"Human Pose and Action Recognition","score":0.05000000074505806,"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/T11975","display_name":"Evolutionary Algorithms and Applications","score":0.02239999920129776,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/recursive-self-improvement","display_name":"recursive self-improvement","score":0.8640000224113464},{"id":"https://openalex.org/keywords/execution-traces","display_name":"execution traces","score":0.4970000088214874},{"id":"https://openalex.org/keywords/evidence-acquisition","display_name":"evidence acquisition","score":0.44699999690055847}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7501000165939331},{"id":"https://openalex.org/C160145156","wikidata":"https://www.wikidata.org/wiki/Q778586","display_name":"Executable","level":2,"score":0.6377000212669373},{"id":"https://openalex.org/C188198153","wikidata":"https://www.wikidata.org/wiki/Q1613840","display_name":"Limiting","level":2,"score":0.4909999966621399},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.42820000648498535},{"id":"https://openalex.org/C107457646","wikidata":"https://www.wikidata.org/wiki/Q207434","display_name":"Human\u2013computer interaction","level":1,"score":0.36970001459121704},{"id":"https://openalex.org/C2776760102","wikidata":"https://www.wikidata.org/wiki/Q5139990","display_name":"Code (set theory)","level":3,"score":0.35899999737739563},{"id":"https://openalex.org/C168167062","wikidata":"https://www.wikidata.org/wiki/Q1117970","display_name":"Component (thermodynamics)","level":2,"score":0.33500000834465027},{"id":"https://openalex.org/C202474056","wikidata":"https://www.wikidata.org/wiki/Q1931635","display_name":"Video tracking","level":3,"score":0.31610000133514404},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.2906000018119812},{"id":"https://openalex.org/C2780791683","wikidata":"https://www.wikidata.org/wiki/Q846785","display_name":"Action (physics)","level":2,"score":0.2825999855995178}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2609.37950","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2609.37950","pdf_url":"https://arxiv.org/pdf/2609.37950","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"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":"other-oa","license_id":"https://openalex.org/licenses/other-oa","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2609.37950","doi":"https://doi.org/10.48550/arxiv.2609.37950","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2609.37950","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"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":false,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2609.37950","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2609.37950","pdf_url":"https://arxiv.org/pdf/2609.37950","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"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":"other-oa","license_id":"https://openalex.org/licenses/other-oa","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7214939883.pdf","grobid_xml":"https://content.openalex.org/works/W7214939883.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W7222307745","https://openalex.org/W7221767287","https://openalex.org/W7213581144","https://openalex.org/W7215081862","https://openalex.org/W7196945490","https://openalex.org/W4414895766","https://openalex.org/W7140077008","https://openalex.org/W4417210844","https://openalex.org/W4416872481","https://openalex.org/W7140472030"],"abstract_inverted_index":{"Video":[0],"understanding":[1,56,124,143,152],"agents":[2,153],"acquire":[3],"evidence":[4,26,90,158],"through":[5,162],"an":[6],"executable":[7],"harness":[8,96,163],"that":[9],"controls":[10],"what":[11],"they":[12,16],"observe":[13],"and":[14,38,104,115,135,160],"how":[15],"use":[17,161],"those":[18],"observations.":[19],"However,":[20],"execution":[21],"traces":[22],"contain":[23],"only":[24],"the":[25,29,40,71,74,92,126,148],"acquired":[27],"by":[28,110],"current":[30],"harness,":[31],"leaving":[32],"competing":[33,80],"explanations":[34,82],"for":[35,42,49,150],"failure":[36,81],"unresolved":[37],"limiting":[39],"basis":[41],"self-improvement.":[43],"We":[44],"introduce":[45],"Video-RSI,":[46],"a":[47,54],"framework":[48],"recursive":[50],"self-improvement":[51],"in":[52,89],"which":[53,106],"video":[55,69,123,142,151],"agent":[57,128],"uses":[58],"its":[59,65],"own":[60,157],"language":[61],"model":[62,72],"to":[63,78,108,154],"revise":[64],"harness.":[66],"Through":[67],"active":[68],"investigation,":[70],"revisits":[73],"original":[75],"training":[76],"videos":[77],"test":[79],"with":[83],"additional":[84],"observations,":[85],"grounding":[86],"proposed":[87],"changes":[88],"beyond":[91],"existing":[93,141],"trace.":[94],"Cost-aware":[95],"evolution":[97],"turns":[98],"these":[99],"diagnoses":[100],"into":[101],"reusable":[102],"revisions":[103,107],"determines":[105],"retain":[109],"considering":[111],"both":[112],"answer":[113],"accuracy":[114,130],"visual":[116],"cost.":[117],"Across":[118],"our":[119],"evaluation":[120],"settings":[121],"on":[122],"benchmarks,":[125],"evolved":[127],"improves":[129],"while":[131],"processing":[132],"fewer":[133],"frames":[134],"achieves":[136],"competitive":[137],"accuracy-efficiency":[138],"trade-offs":[139],"against":[140],"agents.":[144],"These":[145],"results":[146],"demonstrate":[147],"potential":[149],"improve":[155],"their":[156],"acquisition":[159],"evolution.":[164],"Code":[165],"is":[166],"available":[167],"at":[168],"https://github.com/bingjunluo/Video-RSI":[169],".":[170]},"counts_by_year":[],"updated_date":"2026-10-28T10:33:24.094632","created_date":"2026-10-01T00:00:00"}