{"id":"https://openalex.org/W4412377132","doi":"https://doi.org/10.1145/3726302.3730290","title":"Benchmark Granularity and Model Robustness for Image-Text Retrieval: A Reproducibility Study","display_name":"Benchmark Granularity and Model Robustness for Image-Text Retrieval: A Reproducibility Study","publication_year":2025,"publication_date":"2025-07-13","ids":{"openalex":"https://openalex.org/W4412377132","doi":"https://doi.org/10.1145/3726302.3730290"},"language":"en","primary_location":{"id":"doi:10.1145/3726302.3730290","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3726302.3730290","pdf_url":"https://dl.acm.org/doi/pdf/10.1145/3726302.3730290","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 48th International ACM SIGIR Conference on Research and Development in Information Retrieval","raw_type":"proceedings-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://dl.acm.org/doi/pdf/10.1145/3726302.3730290","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5015593025","display_name":"Mariya Hendriksen","orcid":"https://orcid.org/0000-0003-0314-2955"},"institutions":[{"id":"https://openalex.org/I4210135670","display_name":"Amsterdam University of the Arts","ror":"https://ror.org/04dde1554","country_code":"NL","type":"education","lineage":["https://openalex.org/I4210135670"]},{"id":"https://openalex.org/I887064364","display_name":"University of Amsterdam","ror":"https://ror.org/04dkp9463","country_code":"NL","type":"education","lineage":["https://openalex.org/I887064364"]}],"countries":["NL"],"is_corresponding":false,"raw_author_name":"Mariya Hendriksen","raw_affiliation_strings":["University of Amsterdam, Amsterdam, Netherlands"],"raw_orcid":"https://orcid.org/0000-0003-0314-2955","affiliations":[{"raw_affiliation_string":"University of Amsterdam, Amsterdam, Netherlands","institution_ids":["https://openalex.org/I4210135670","https://openalex.org/I887064364"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100450409","display_name":"Shuo Zhang","orcid":"https://orcid.org/0000-0003-3179-4125"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Shuo Zhang","raw_affiliation_strings":["Bloomberg, London, United Kingdom"],"raw_orcid":"https://orcid.org/0000-0003-3179-4125","affiliations":[{"raw_affiliation_string":"Bloomberg, London, United Kingdom","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5065148077","display_name":"Ridho Reinanda","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ridho Reinanda","raw_affiliation_strings":["Bloomberg, London, United Kingdom"],"raw_orcid":"https://orcid.org/0009-0008-5316-2683","affiliations":[{"raw_affiliation_string":"Bloomberg, London, United Kingdom","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5087929673","display_name":"Mohamed Yahya","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Mohamed Yahya","raw_affiliation_strings":["Bloomberg, London, United Kingdom"],"raw_orcid":"https://orcid.org/0009-0002-1260-8059","affiliations":[{"raw_affiliation_string":"Bloomberg, London, United Kingdom","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5031940626","display_name":"Edgar Meij","orcid":"https://orcid.org/0000-0003-0516-3688"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Edgar Meij","raw_affiliation_strings":["Bloomberg, London, United Kingdom"],"raw_orcid":"https://orcid.org/0000-0003-0516-3688","affiliations":[{"raw_affiliation_string":"Bloomberg, London, United Kingdom","institution_ids":[]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5031439294","display_name":"Maarten de Rijke","orcid":"https://orcid.org/0000-0002-1086-0202"},"institutions":[{"id":"https://openalex.org/I4210135670","display_name":"Amsterdam University of the Arts","ror":"https://ror.org/04dde1554","country_code":"NL","type":"education","lineage":["https://openalex.org/I4210135670"]},{"id":"https://openalex.org/I887064364","display_name":"University of Amsterdam","ror":"https://ror.org/04dkp9463","country_code":"NL","type":"education","lineage":["https://openalex.org/I887064364"]}],"countries":["NL"],"is_corresponding":false,"raw_author_name":"Maarten de Rijke","raw_affiliation_strings":["University of Amsterdam, Amsterdam, Netherlands"],"raw_orcid":"https://orcid.org/0000-0002-1086-0202","affiliations":[{"raw_affiliation_string":"University of Amsterdam, Amsterdam, Netherlands","institution_ids":["https://openalex.org/I4210135670","https://openalex.org/I887064364"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":6,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.9349,"has_fulltext":true,"cited_by_count":1,"citation_normalized_percentile":{"value":0.76457463,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":95,"max":98},"biblio":{"volume":null,"issue":null,"first_page":"3183","last_page":"3193"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11714","display_name":"Multimodal Machine Learning Applications","score":1.0,"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":1.0,"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/T10028","display_name":"Topic Modeling","score":0.9977999925613403,"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/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.9976000189781189,"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/robustness","display_name":"Robustness (evolution)","score":0.8100546002388},{"id":"https://openalex.org/keywords/granularity","display_name":"Granularity","score":0.7413824200630188},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.7271198034286499},{"id":"https://openalex.org/keywords/reproducibility","display_name":"Reproducibility","score":0.5971835255622864},{"id":"https://openalex.org/keywords/benchmark","display_name":"Benchmark (surveying)","score":0.5644300580024719},{"id":"https://openalex.org/keywords/image-retrieval","display_name":"Image retrieval","score":0.4509054720401764},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.4375688433647156},{"id":"https://openalex.org/keywords/data-mining","display_name":"Data mining","score":0.38714125752449036},{"id":"https://openalex.org/keywords/pattern-recognition","display_name":"Pattern recognition (psychology)","score":0.3763670325279236},{"id":"https://openalex.org/keywords/information-retrieval","display_name":"Information retrieval","score":0.369992196559906},{"id":"https://openalex.org/keywords/image","display_name":"Image (mathematics)","score":0.3122518062591553},{"id":"https://openalex.org/keywords/mathematics","display_name":"Mathematics","score":0.1784747838973999},{"id":"https://openalex.org/keywords/programming-language","display_name":"Programming language","score":0.11786073446273804},{"id":"https://openalex.org/keywords/statistics","display_name":"Statistics","score":0.11760121583938599}],"concepts":[{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.8100546002388},{"id":"https://openalex.org/C177774035","wikidata":"https://www.wikidata.org/wiki/Q1246948","display_name":"Granularity","level":2,"score":0.7413824200630188},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7271198034286499},{"id":"https://openalex.org/C9893847","wikidata":"https://www.wikidata.org/wiki/Q1425625","display_name":"Reproducibility","level":2,"score":0.5971835255622864},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.5644300580024719},{"id":"https://openalex.org/C1667742","wikidata":"https://www.wikidata.org/wiki/Q10927554","display_name":"Image retrieval","level":3,"score":0.4509054720401764},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4375688433647156},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.38714125752449036},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.3763670325279236},{"id":"https://openalex.org/C23123220","wikidata":"https://www.wikidata.org/wiki/Q816826","display_name":"Information retrieval","level":1,"score":0.369992196559906},{"id":"https://openalex.org/C115961682","wikidata":"https://www.wikidata.org/wiki/Q860623","display_name":"Image (mathematics)","level":2,"score":0.3122518062591553},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.1784747838973999},{"id":"https://openalex.org/C199360897","wikidata":"https://www.wikidata.org/wiki/Q9143","display_name":"Programming language","level":1,"score":0.11786073446273804},{"id":"https://openalex.org/C105795698","wikidata":"https://www.wikidata.org/wiki/Q12483","display_name":"Statistics","level":1,"score":0.11760121583938599},{"id":"https://openalex.org/C55493867","wikidata":"https://www.wikidata.org/wiki/Q7094","display_name":"Biochemistry","level":1,"score":0.0},{"id":"https://openalex.org/C13280743","wikidata":"https://www.wikidata.org/wiki/Q131089","display_name":"Geodesy","level":1,"score":0.0},{"id":"https://openalex.org/C104317684","wikidata":"https://www.wikidata.org/wiki/Q7187","display_name":"Gene","level":2,"score":0.0},{"id":"https://openalex.org/C185592680","wikidata":"https://www.wikidata.org/wiki/Q2329","display_name":"Chemistry","level":0,"score":0.0},{"id":"https://openalex.org/C205649164","wikidata":"https://www.wikidata.org/wiki/Q1071","display_name":"Geography","level":0,"score":0.0}],"mesh":[],"locations_count":2,"locations":[{"id":"doi:10.1145/3726302.3730290","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3726302.3730290","pdf_url":"https://dl.acm.org/doi/pdf/10.1145/3726302.3730290","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 48th International ACM SIGIR Conference on Research and Development in Information Retrieval","raw_type":"proceedings-article"},{"id":"pmh:oai:dare.uva.nl:openaire/3bf768e2-2e4c-4aff-b569-18f572dec97f","is_oa":true,"landing_page_url":"https://handle.uba.uva.nl/personal/pure/en/publications/benchmark-granularity-and-model-robustness-for-imagetext-retrieval(3bf768e2-2e4c-4aff-b569-18f572dec97f).html","pdf_url":null,"source":{"id":"https://openalex.org/S4306400088","display_name":"UvA-DARE (University of Amsterdam)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I887064364","host_organization_name":"University of Amsterdam","host_organization_lineage":["https://openalex.org/I887064364"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Hendriksen, M, Zhang, S, Reinanda, R, Yahya, M, Meij, E & de Rijke, M 2025, Benchmark Granularity and Model Robustness for Image-Text Retrieval : A Reproducibility Study. in SIGIR '25 : Proceedings of the 48th International ACM SIGIR Conference on Research and Development in Information Retrieval : July 13-18, 2025, Padua, Italy. Association for Computing Machinery, New York, NY, pp. 3183-3193, 48th International ACM SIGIR Conference on Research and Development in Information Retrieval, SIGIR 2025, Padua, Italy, 13/07/25. https://doi.org/10.1145/3726302.3730290","raw_type":"info:eu-repo/semantics/publishedVersion"}],"best_oa_location":{"id":"doi:10.1145/3726302.3730290","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3726302.3730290","pdf_url":"https://dl.acm.org/doi/pdf/10.1145/3726302.3730290","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 48th International ACM SIGIR Conference on Research and Development in Information Retrieval","raw_type":"proceedings-article"},"sustainable_development_goals":[{"display_name":"Gender equality","id":"https://metadata.un.org/sdg/5","score":0.5600000023841858}],"awards":[{"id":"https://openalex.org/G2289169492","display_name":"Fairness and Intersectional Non-Discrimination in Human Recommendation","funder_award_id":"101070212","funder_id":"https://openalex.org/F4320320300","funder_display_name":"European Commission"},{"id":"https://openalex.org/G5550786340","display_name":null,"funder_award_id":"024.004.022","funder_id":"https://openalex.org/F4320321800","funder_display_name":"Nederlandse Organisatie voor Wetenschappelijk Onderzoek"},{"id":"https://openalex.org/G6280615682","display_name":null,"funder_award_id":"NWA.1389.20.183","funder_id":"https://openalex.org/F4320321800","funder_display_name":"Nederlandse Organisatie voor Wetenschappelijk Onderzoek"}],"funders":[{"id":"https://openalex.org/F4320320300","display_name":"European Commission","ror":"https://ror.org/00k4n6c32"},{"id":"https://openalex.org/F4320321800","display_name":"Nederlandse Organisatie voor Wetenschappelijk Onderzoek","ror":"https://ror.org/04jsz6e67"},{"id":"https://openalex.org/F4320334322","display_name":"HORIZON EUROPE Framework Programme","ror":null}],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4412377132.pdf","grobid_xml":"https://content.openalex.org/works/W4412377132.grobid-xml"},"referenced_works_count":54,"referenced_works":["https://openalex.org/W1905882502","https://openalex.org/W2069065514","https://openalex.org/W2185175083","https://openalex.org/W2396566415","https://openalex.org/W2481240925","https://openalex.org/W2546696630","https://openalex.org/W2604272474","https://openalex.org/W2619383789","https://openalex.org/W2783914869","https://openalex.org/W2886641317","https://openalex.org/W2954182683","https://openalex.org/W2960416371","https://openalex.org/W2962964995","https://openalex.org/W2962995403","https://openalex.org/W2963389687","https://openalex.org/W2966491090","https://openalex.org/W2967593235","https://openalex.org/W2969862959","https://openalex.org/W2970231061","https://openalex.org/W2996851481","https://openalex.org/W2998356391","https://openalex.org/W3000155959","https://openalex.org/W3035965352","https://openalex.org/W3099878876","https://openalex.org/W3106681699","https://openalex.org/W3135053698","https://openalex.org/W3135367836","https://openalex.org/W3153594491","https://openalex.org/W3166396011","https://openalex.org/W3173681001","https://openalex.org/W3174169056","https://openalex.org/W3176549216","https://openalex.org/W3200686557","https://openalex.org/W3213100861","https://openalex.org/W4212804468","https://openalex.org/W4224980447","https://openalex.org/W4225166070","https://openalex.org/W4229907684","https://openalex.org/W4283317927","https://openalex.org/W4288262986","https://openalex.org/W4293567540","https://openalex.org/W4301409532","https://openalex.org/W4312980231","https://openalex.org/W4327644550","https://openalex.org/W4365799947","https://openalex.org/W4366851068","https://openalex.org/W4385573584","https://openalex.org/W4387846590","https://openalex.org/W4400527475","https://openalex.org/W6600339457","https://openalex.org/W6600424091","https://openalex.org/W6600459194","https://openalex.org/W6600617704","https://openalex.org/W6949549905"],"related_works":["https://openalex.org/W2413717610","https://openalex.org/W1973270181","https://openalex.org/W2417696084","https://openalex.org/W2368782778","https://openalex.org/W2087830269","https://openalex.org/W3106281778","https://openalex.org/W2931688134","https://openalex.org/W4324138256","https://openalex.org/W3134060498","https://openalex.org/W2003643616"],"abstract_inverted_index":{"Image-Text":[0],"Retrieval":[1],"(ITR)":[2],"systems":[3],"are":[4],"central":[5],"to":[6,31,102,151],"multimodal":[7],"information":[8],"access,":[9],"with":[10],"Vision-Language":[11],"Models":[12],"(VLMs)":[13],"showing":[14],"strong":[15],"performance":[16,58],"on":[17,25,177],"standard":[18,73],"benchmarks.":[19],"However,":[20],"these":[21],"benchmarks":[22,74],"predominantly":[23],"rely":[24],"coarse-grained":[26],"annotations,":[27],"limiting":[28],"their":[29,78],"ability":[30],"reveal":[32],"how":[33,50],"models":[34,176],"would":[35],"perform":[36],"under":[37],"real-world":[38],"conditions,":[39],"where":[40,93],"query":[41,54],"granularity":[42,52,166],"varies.":[43],"Motivated":[44],"by":[45],"this":[46],"gap,":[47],"we":[48,81,94,109],"examine":[49],"dataset":[51],"and":[53,59,69,77,115,160,167,170],"perturbations":[55,114,122],"affect":[56],"retrieval":[57],"robustness":[60,159],"across":[61],"four":[62],"architecturally":[63],"diverse":[64],"VLMs":[65],"(ALIGN,":[66],"AltCLIP,":[67],"CLIP,":[68],"GroupViT).":[70],"Using":[71],"both":[72],"(MS-COCO,":[75],"Flickr30k)":[76],"fine-grained":[79],"variants,":[80],"show":[82],"that":[83,120],"richer":[84],"captions":[85],"consistently":[86],"enhance":[87],"retrieval,":[88,131],"especially":[89],"in":[90,104,157],"text-to-image":[91],"tasks,":[92],"observe":[95],"an":[96],"average":[97],"improvement":[98],"of":[99,113,148,174,179],"16.23%,":[100],"compared":[101],"6.44%":[103],"image-to-text.":[105],"To":[106],"assess":[107],"robustness,":[108],"introduce":[110],"a":[111,141,161],"taxonomy":[112],"conduct":[116],"extensive":[117],"experiments,":[118],"revealing":[119],"while":[121],"typically":[123],"degrade":[124],"performance,":[125],"they":[126],"can":[127],"also":[128],"unexpectedly":[129],"improve":[130],"exposing":[132],"nuanced":[133],"model":[134,149,158],"behaviors.":[135],"Notably,":[136],"word":[137],"order":[138],"emerges":[139],"as":[140],"critical":[142],"factor":[143],"-":[144],"contradicting":[145],"prior":[146],"assumptions":[147],"insensitivity":[150],"it.":[152],"Our":[153],"results":[154],"highlight":[155],"variation":[156],"dataset-dependent":[162],"relationship":[163],"between":[164],"caption":[165],"perturbation":[168],"sensitivity":[169],"emphasize":[171],"the":[172],"necessity":[173],"evaluating":[175],"datasets":[178],"varying":[180],"granularity.":[181]},"counts_by_year":[{"year":2026,"cited_by_count":1}],"updated_date":"2026-06-11T09:08:48.828518","created_date":"2025-10-10T00:00:00"}