{"id":"https://openalex.org/W7155558949","doi":"https://doi.org/10.48550/arxiv.2604.21523","title":"Seeing Isn't Believing: Uncovering Blind Spots in Evaluator Vision-Language Models","display_name":"Seeing Isn't Believing: Uncovering Blind Spots in Evaluator Vision-Language Models","publication_year":2026,"publication_date":"2026-04-23","ids":{"openalex":"https://openalex.org/W7155558949","doi":"https://doi.org/10.48550/arxiv.2604.21523"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2604.21523","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.21523","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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"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":"https://doi.org/10.48550/arxiv.2604.21523","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5093817486","display_name":"Mohammed Safi Ur Rahman Khan","orcid":null},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Khan, Mohammed Safi Ur Rahman","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5126337991","display_name":"Sanjay Suryanarayanan","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Suryanarayanan, Sanjay","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101332925","display_name":"Tushar Anand","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Anand, Tushar","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5134551708","display_name":"Mitesh M. Khapra","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Khapra, Mitesh M.","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":4,"corresponding_author_ids":["https://openalex.org/A5093817486"],"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":0.9904999732971191,"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":0.9904999732971191,"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":0.003000000026077032,"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":0.0010999999940395355,"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/pairwise-comparison","display_name":"Pairwise comparison","score":0.6384000182151794},{"id":"https://openalex.org/keywords/benchmarking","display_name":"Benchmarking","score":0.5758000016212463},{"id":"https://openalex.org/keywords/benchmark","display_name":"Benchmark (surveying)","score":0.5192999839782715},{"id":"https://openalex.org/keywords/reliability","display_name":"Reliability (semiconductor)","score":0.5105999708175659},{"id":"https://openalex.org/keywords/unobservable","display_name":"Unobservable","score":0.41940000653266907},{"id":"https://openalex.org/keywords/hallucinating","display_name":"Hallucinating","score":0.41530001163482666},{"id":"https://openalex.org/keywords/data-quality","display_name":"Data quality","score":0.352400004863739},{"id":"https://openalex.org/keywords/visual-inspection","display_name":"Visual inspection","score":0.34369999170303345}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6697999835014343},{"id":"https://openalex.org/C184898388","wikidata":"https://www.wikidata.org/wiki/Q1435712","display_name":"Pairwise comparison","level":2,"score":0.6384000182151794},{"id":"https://openalex.org/C86251818","wikidata":"https://www.wikidata.org/wiki/Q816754","display_name":"Benchmarking","level":2,"score":0.5758000016212463},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.521399974822998},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.5192999839782715},{"id":"https://openalex.org/C43214815","wikidata":"https://www.wikidata.org/wiki/Q7310987","display_name":"Reliability (semiconductor)","level":3,"score":0.5105999708175659},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.4706999957561493},{"id":"https://openalex.org/C2780695315","wikidata":"https://www.wikidata.org/wiki/Q3799040","display_name":"Unobservable","level":2,"score":0.41940000653266907},{"id":"https://openalex.org/C2911011789","wikidata":"https://www.wikidata.org/wiki/Q130741","display_name":"Hallucinating","level":2,"score":0.41530001163482666},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.3955000042915344},{"id":"https://openalex.org/C24756922","wikidata":"https://www.wikidata.org/wiki/Q1757694","display_name":"Data quality","level":3,"score":0.352400004863739},{"id":"https://openalex.org/C168820333","wikidata":"https://www.wikidata.org/wiki/Q448889","display_name":"Visual inspection","level":2,"score":0.34369999170303345},{"id":"https://openalex.org/C2779530757","wikidata":"https://www.wikidata.org/wiki/Q1207505","display_name":"Quality (philosophy)","level":2,"score":0.3425999879837036},{"id":"https://openalex.org/C2781238097","wikidata":"https://www.wikidata.org/wiki/Q175026","display_name":"Object (grammar)","level":2,"score":0.3334999978542328},{"id":"https://openalex.org/C107673813","wikidata":"https://www.wikidata.org/wiki/Q812534","display_name":"Bayesian probability","level":2,"score":0.310699999332428},{"id":"https://openalex.org/C43126263","wikidata":"https://www.wikidata.org/wiki/Q128751","display_name":"Source code","level":2,"score":0.3059000074863434},{"id":"https://openalex.org/C3020001037","wikidata":"https://www.wikidata.org/wiki/Q836575","display_name":"Quality assessment","level":3,"score":0.30480000376701355},{"id":"https://openalex.org/C2776760102","wikidata":"https://www.wikidata.org/wiki/Q5139990","display_name":"Code (set theory)","level":3,"score":0.29440000653266907},{"id":"https://openalex.org/C31972630","wikidata":"https://www.wikidata.org/wiki/Q844240","display_name":"Computer vision","level":1,"score":0.2921000123023987},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.2888999879360199},{"id":"https://openalex.org/C16910744","wikidata":"https://www.wikidata.org/wiki/Q7705759","display_name":"Test data","level":2,"score":0.26109999418258667},{"id":"https://openalex.org/C3018395757","wikidata":"https://www.wikidata.org/wiki/Q1379672","display_name":"Evaluation methods","level":2,"score":0.2533999979496002}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2604.21523","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.21523","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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"article"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2604.21523","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.21523","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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Large":[0],"Vision-Language":[1],"Models":[2],"(VLMs)":[3],"are":[4,155],"increasingly":[5],"used":[6],"to":[7,136,158],"evaluate":[8,45,110],"outputs":[9,139],"of":[10,33,48,100,181],"other":[11],"models,":[12],"for":[13,88,191],"image-to-text":[14],"(I2T)":[15],"tasks":[16],"such":[17],"as":[18],"visual":[19,77],"question":[20],"answering,":[21],"and":[22,54,76,119,151,154,185,193,197],"text-to-image":[23],"(T2I)":[24],"generation":[25],"tasks.":[26,56],"Despite":[27],"this":[28,41],"growing":[29],"reliance,":[30],"the":[31,46,163,178],"reliability":[32,47],"these":[34,89],"Evaluator":[35,49,83,183],"VLMs":[36,50,84,113,184],"remains":[37],"under":[38],"explored.":[39],"In":[40],"work,":[42],"we":[43,109],"systematically":[44],"across":[51],"both":[52],"I2T":[53],"T2I":[55],"We":[57],"introduce":[58],"targeted":[59],"perturbations":[60,80],"that":[61,125,161],"degrade":[62],"output":[63],"quality":[64,90],"along":[65],"key":[66],"error":[67],"dimensions,":[68,108],"including":[69],"object":[70],"hallucinations,":[71],"spatial":[72,152],"reasoning,":[73],"factual":[74],"grounding,":[75],"fidelity.":[78],"These":[79,175],"test":[81],"whether":[82],"can":[85],"reliably":[86],"account":[87],"degrading":[91],"errors":[92],"in":[93,141,188],"their":[94,189],"evaluations.":[95],"Using":[96],"a":[97],"comprehensive":[98],"benchmark":[99],"over":[101],"4000":[102],"perturbed":[103,138],"instances":[104],"spanning":[105],"40":[106],"perturbation":[107],"4":[111],"prominent":[112],"using":[114],"single-answer":[115],"scoring,":[116],"pairwise":[117],"comparison,":[118],"reference-guided":[120],"paradigms.":[121],"Our":[122],"findings":[123],"reveal":[124],"current":[126,182],"VLM":[127],"evaluators":[128],"exhibit":[129],"substantial":[130],"blind":[131],"spots:":[132],"they":[133],"often":[134,156],"fail":[135],"detect":[137],"-":[140],"some":[142],"cases":[143],"exceeding":[144],"50%,":[145],"struggle":[146],"particularly":[147],"with":[148],"fine-grained":[149],"compositional":[150],"errors,":[153],"insensitive":[157],"hallucinated":[159],"content":[160],"contradicts":[162],"input":[164],"image.":[165],"Pairwise":[166],"comparison":[167],"proves":[168],"more":[169],"reliable,":[170],"though":[171],"failure":[172],"rates":[173],"persist.":[174],"results":[176],"highlight":[177],"unreliable":[179],"nature":[180],"urge":[186],"caution":[187],"deployment":[190],"benchmarking":[192],"development":[194],"decisions.":[195],"Code":[196],"data":[198],"have":[199],"been":[200],"made":[201],"publicly":[202],"available.":[203]},"counts_by_year":[],"updated_date":"2026-04-25T06:06:54.107920","created_date":"2026-04-25T00:00:00"}