{"id":"https://openalex.org/W7140090746","doi":"https://doi.org/10.48550/arxiv.2603.19775","title":"Evaluating Image Editing with LLMs: A Comprehensive Benchmark and Intermediate-Layer Probing Approach","display_name":"Evaluating Image Editing with LLMs: A Comprehensive Benchmark and Intermediate-Layer Probing Approach","publication_year":2026,"publication_date":"2026-03-20","ids":{"openalex":"https://openalex.org/W7140090746","doi":"https://doi.org/10.48550/arxiv.2603.19775"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2603.19775","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.19775","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":false,"raw_source_name":null,"raw_type":"Preprint"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2603.19775","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5121210964","display_name":"Shiqi Gao","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Gao, Shiqi","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5130407489","display_name":"Zitong Xu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xu, Zitong","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5130368939","display_name":"Kang Fu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Fu, Kang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5130370679","display_name":"Huiyu Duan","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Duan, Huiyu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5130382426","display_name":"Xiongkuo Min","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Min, Xiongkuo","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5130360645","display_name":"Jia Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"wang, Jia","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"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/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.44119998812675476,"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/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.44119998812675476,"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/T12377","display_name":"Digital Humanities and Scholarship","score":0.09319999814033508,"subfield":{"id":"https://openalex.org/subfields/1208","display_name":"Literature and Literary Theory"},"field":{"id":"https://openalex.org/fields/12","display_name":"Arts and Humanities"},"domain":{"id":"https://openalex.org/domains/2","display_name":"Social Sciences"}},{"id":"https://openalex.org/T11714","display_name":"Multimodal Machine Learning Applications","score":0.03269999846816063,"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/image-editing","display_name":"Image editing","score":0.8287000060081482},{"id":"https://openalex.org/keywords/benchmark","display_name":"Benchmark (surveying)","score":0.7491999864578247},{"id":"https://openalex.org/keywords/perception","display_name":"Perception","score":0.5490000247955322},{"id":"https://openalex.org/keywords/image","display_name":"Image (mathematics)","score":0.4828999936580658},{"id":"https://openalex.org/keywords/quality","display_name":"Quality (philosophy)","score":0.3993000090122223},{"id":"https://openalex.org/keywords/image-quality","display_name":"Image quality","score":0.323199987411499}],"concepts":[{"id":"https://openalex.org/C2776674983","wikidata":"https://www.wikidata.org/wiki/Q545981","display_name":"Image editing","level":3,"score":0.8287000060081482},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7842000126838684},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.7491999864578247},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5947999954223633},{"id":"https://openalex.org/C26760741","wikidata":"https://www.wikidata.org/wiki/Q160402","display_name":"Perception","level":2,"score":0.5490000247955322},{"id":"https://openalex.org/C115961682","wikidata":"https://www.wikidata.org/wiki/Q860623","display_name":"Image (mathematics)","level":2,"score":0.4828999936580658},{"id":"https://openalex.org/C2779530757","wikidata":"https://www.wikidata.org/wiki/Q1207505","display_name":"Quality (philosophy)","level":2,"score":0.3993000090122223},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.35370001196861267},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.32899999618530273},{"id":"https://openalex.org/C31972630","wikidata":"https://www.wikidata.org/wiki/Q844240","display_name":"Computer vision","level":1,"score":0.3287000060081482},{"id":"https://openalex.org/C55020928","wikidata":"https://www.wikidata.org/wiki/Q3813865","display_name":"Image quality","level":3,"score":0.323199987411499},{"id":"https://openalex.org/C23123220","wikidata":"https://www.wikidata.org/wiki/Q816826","display_name":"Information retrieval","level":1,"score":0.3043999969959259},{"id":"https://openalex.org/C62897895","wikidata":"https://www.wikidata.org/wiki/Q1915482","display_name":"Mean opinion score","level":3,"score":0.28870001435279846},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.28189998865127563},{"id":"https://openalex.org/C132964779","wikidata":"https://www.wikidata.org/wiki/Q2110223","display_name":"Raw data","level":2,"score":0.27230000495910645},{"id":"https://openalex.org/C2780310081","wikidata":"https://www.wikidata.org/wiki/Q1154312","display_name":"Video editing","level":2,"score":0.26739999651908875},{"id":"https://openalex.org/C184337299","wikidata":"https://www.wikidata.org/wiki/Q1437428","display_name":"Semantics (computer science)","level":2,"score":0.2540999948978424}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2603.19775","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.19775","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":"doi:10.48550/arxiv.2603.19775","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.19775","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":false,"raw_source_name":null,"raw_type":"Preprint"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Evaluating":[0],"text-guided":[1,62,223],"image":[2,26,63,224],"editing":[3,64,74,79,120,138,176,197,225],"(TIE)":[4],"methods":[5],"remains":[6],"a":[7,56,213],"challenging":[8],"problem,":[9],"as":[10],"reliable":[11,93,217],"assessment":[12],"should":[13],"simultaneously":[14],"consider":[15],"perceptual":[16,48,118,171],"quality,":[17,119],"alignment":[18,204],"with":[19,46,73,193,205],"textual":[20],"instructions,":[21,177],"and":[22,41,122,170,178,210,218],"preservation":[23],"of":[24,61,68,143,147,161,222],"original":[25],"content.":[27],"Despite":[28],"rapid":[29],"progress":[30],"in":[31,39],"TIE":[32,89],"models,":[33],"existing":[34],"evaluation":[35,60,116,188,221],"benchmarks":[36],"remain":[37],"limited":[38,191],"scale":[40],"often":[42],"show":[43,190],"weak":[44],"correlation":[45,192],"human":[47,194,206],"judgments.":[49],"In":[50],"this":[51],"work,":[52],"we":[53,129],"introduce":[54],"TIEdit,":[55],"benchmark":[57,127],"for":[58,215],"systematic":[59],"methods.":[65,226],"TIEdit":[66,209],"consists":[67],"512":[69],"source":[70,174],"images":[71,84],"paired":[72],"prompts":[75],"across":[76,114],"eight":[77],"representative":[78],"tasks,":[80,198],"producing":[81],"5,120":[82],"edited":[83,179],"generated":[85],"by":[86],"ten":[87],"state-of-the-art":[88],"models.":[90],"To":[91],"obtain":[92],"subjective":[94,104],"ratings,":[95,105],"20":[96],"experts":[97],"are":[98],"recruited":[99],"to":[100,166],"produce":[101],"307,200":[102],"raw":[103],"which":[106],"accumulates":[107],"into":[108],"15,360":[109],"mean":[110],"opinion":[111],"scores":[112],"(MOSs)":[113],"three":[115],"dimensions:":[117],"alignment,":[121],"content":[123],"preservation.":[124],"Beyond":[125],"the":[126],"itself,":[128],"further":[130],"propose":[131],"EditProbe,":[132],"an":[133],"LLM-based":[134],"evaluator":[135],"that":[136,184],"estimates":[137],"quality":[139],"via":[140],"intermediate-layer":[141],"probing":[142],"hidden":[144],"representations.":[145],"Instead":[146],"relying":[148],"solely":[149],"on":[150,196],"final":[151],"model":[152],"outputs,":[153],"EditProbe":[154,200,211],"extracts":[155],"informative":[156],"representations":[157],"from":[158],"intermediate":[159],"layers":[160],"multimodal":[162],"large":[163],"language":[164],"models":[165],"better":[167],"capture":[168],"semantic":[169],"relationships":[172],"between":[173],"images,":[175],"results.":[180],"Experimental":[181],"results":[182],"demonstrate":[183],"widely":[185],"used":[186],"automatic":[187],"metrics":[189],"judgments":[195],"while":[199],"achieves":[201],"substantially":[202],"stronger":[203],"perception.":[207],"Together,":[208],"provide":[212],"foundation":[214],"more":[216],"perceptually":[219],"aligned":[220]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-03-24T00:00:00"}