{"id":"https://openalex.org/W4414929902","doi":"https://doi.org/10.1109/iccv51701.2025.02262","title":"GenHancer: Imperfect Generative Models are Secretly Strong Vision-Centric Enhancers","display_name":"GenHancer: Imperfect Generative Models are Secretly Strong Vision-Centric Enhancers","publication_year":2025,"publication_date":"2025-10-19","ids":{"openalex":"https://openalex.org/W4414929902","doi":"https://doi.org/10.1109/iccv51701.2025.02262"},"language":"en","primary_location":{"id":"doi:10.1109/iccv51701.2025.02262","is_oa":false,"landing_page_url":"https://doi.org/10.1109/iccv51701.2025.02262","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2025 IEEE/CVF International Conference on Computer Vision (ICCV)","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["arxiv","crossref","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2503.19480","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5101545102","display_name":"Shijie Ma","orcid":"https://orcid.org/0000-0003-4423-0142"},"institutions":[{"id":"https://openalex.org/I2250653659","display_name":"Tencent (China)","ror":"https://ror.org/00hhjss72","country_code":"CN","type":"company","lineage":["https://openalex.org/I2250653659"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Shijie Ma","raw_affiliation_strings":["Tencent PCG,ARC Lab"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Tencent PCG,ARC Lab","institution_ids":["https://openalex.org/I2250653659"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5062830848","display_name":"Yuying Ge","orcid":"https://orcid.org/0000-0001-5818-2589"},"institutions":[{"id":"https://openalex.org/I2250653659","display_name":"Tencent (China)","ror":"https://ror.org/00hhjss72","country_code":"CN","type":"company","lineage":["https://openalex.org/I2250653659"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yuying Ge","raw_affiliation_strings":["Tencent PCG,ARC Lab"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Tencent PCG,ARC Lab","institution_ids":["https://openalex.org/I2250653659"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5089778019","display_name":"Teng Wang","orcid":"https://orcid.org/0000-0003-3729-0139"},"institutions":[{"id":"https://openalex.org/I2250653659","display_name":"Tencent (China)","ror":"https://ror.org/00hhjss72","country_code":"CN","type":"company","lineage":["https://openalex.org/I2250653659"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Teng Wang","raw_affiliation_strings":["Tencent PCG,ARC Lab"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Tencent PCG,ARC Lab","institution_ids":["https://openalex.org/I2250653659"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101501426","display_name":"Yuxin Guo","orcid":"https://orcid.org/0000-0001-7459-9456"},"institutions":[{"id":"https://openalex.org/I2250653659","display_name":"Tencent (China)","ror":"https://ror.org/00hhjss72","country_code":"CN","type":"company","lineage":["https://openalex.org/I2250653659"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yuxin Guo","raw_affiliation_strings":["Tencent PCG,ARC Lab"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Tencent PCG,ARC Lab","institution_ids":["https://openalex.org/I2250653659"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5047802386","display_name":"Yixiao Ge","orcid":"https://orcid.org/0000-0002-5351-5329"},"institutions":[{"id":"https://openalex.org/I2250653659","display_name":"Tencent (China)","ror":"https://ror.org/00hhjss72","country_code":"CN","type":"company","lineage":["https://openalex.org/I2250653659"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yixiao Ge","raw_affiliation_strings":["Tencent PCG,ARC Lab"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Tencent PCG,ARC Lab","institution_ids":["https://openalex.org/I2250653659"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5088475144","display_name":"Ying Shan","orcid":"https://orcid.org/0000-0003-3253-3197"},"institutions":[{"id":"https://openalex.org/I2250653659","display_name":"Tencent (China)","ror":"https://ror.org/00hhjss72","country_code":"CN","type":"company","lineage":["https://openalex.org/I2250653659"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Ying Shan","raw_affiliation_strings":["Tencent PCG,ARC Lab"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Tencent PCG,ARC Lab","institution_ids":["https://openalex.org/I2250653659"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I2250653659"],"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":"24402","last_page":"24412"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.8198999762535095,"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"}},"topics":[{"id":"https://openalex.org/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.8198999762535095,"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/T10627","display_name":"Advanced Image and Video Retrieval Techniques","score":0.7706999778747559,"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/discriminative-model","display_name":"Discriminative model","score":0.8428000211715698},{"id":"https://openalex.org/keywords/generative-grammar","display_name":"Generative grammar","score":0.766700029373169},{"id":"https://openalex.org/keywords/imperfect","display_name":"Imperfect","score":0.6826000213623047},{"id":"https://openalex.org/keywords/generative-model","display_name":"Generative model","score":0.5885999798774719},{"id":"https://openalex.org/keywords/representation","display_name":"Representation (politics)","score":0.5654000043869019},{"id":"https://openalex.org/keywords/generative-design","display_name":"Generative Design","score":0.3700000047683716}],"concepts":[{"id":"https://openalex.org/C97931131","wikidata":"https://www.wikidata.org/wiki/Q5282087","display_name":"Discriminative model","level":2,"score":0.8428000211715698},{"id":"https://openalex.org/C39890363","wikidata":"https://www.wikidata.org/wiki/Q36108","display_name":"Generative grammar","level":2,"score":0.766700029373169},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7168999910354614},{"id":"https://openalex.org/C2780310539","wikidata":"https://www.wikidata.org/wiki/Q12547192","display_name":"Imperfect","level":2,"score":0.6826000213623047},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.6068000197410583},{"id":"https://openalex.org/C167966045","wikidata":"https://www.wikidata.org/wiki/Q5532625","display_name":"Generative model","level":3,"score":0.5885999798774719},{"id":"https://openalex.org/C2776359362","wikidata":"https://www.wikidata.org/wiki/Q2145286","display_name":"Representation (politics)","level":3,"score":0.5654000043869019},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.45170000195503235},{"id":"https://openalex.org/C184408114","wikidata":"https://www.wikidata.org/wiki/Q1502022","display_name":"Generative Design","level":3,"score":0.3700000047683716},{"id":"https://openalex.org/C125411270","wikidata":"https://www.wikidata.org/wiki/Q18653","display_name":"Encoding (memory)","level":2,"score":0.33730000257492065},{"id":"https://openalex.org/C59404180","wikidata":"https://www.wikidata.org/wiki/Q17013334","display_name":"Feature learning","level":2,"score":0.3077000081539154},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.2840000092983246},{"id":"https://openalex.org/C2777655017","wikidata":"https://www.wikidata.org/wiki/Q1501161","display_name":"Toolbox","level":2,"score":0.2605000138282776},{"id":"https://openalex.org/C2776502983","wikidata":"https://www.wikidata.org/wiki/Q690182","display_name":"Contrast (vision)","level":2,"score":0.2603999972343445},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.2554999887943268}],"mesh":[],"locations_count":3,"locations":[{"id":"doi:10.1109/iccv51701.2025.02262","is_oa":false,"landing_page_url":"https://doi.org/10.1109/iccv51701.2025.02262","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2025 IEEE/CVF International Conference on Computer Vision (ICCV)","raw_type":"proceedings-article"},{"id":"pmh:oai:arXiv.org:2503.19480","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2503.19480","pdf_url":"https://arxiv.org/pdf/2503.19480","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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2503.19480","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2503.19480","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":"pmh:oai:arXiv.org:2503.19480","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2503.19480","pdf_url":"https://arxiv.org/pdf/2503.19480","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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"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":{"The":[0,64,210],"synergy":[1],"between":[2],"generative":[3,31,73],"and":[4,171,229],"discriminative":[5,11],"models":[6,32,74,221,228],"receives":[7],"growing":[8],"attention.":[9],"While":[10],"Contrastive":[12],"Language-Image":[13],"Pre-Training":[14],"(CLIP)":[15],"excels":[16],"in":[17,67],"high-level":[18],"semantics,":[19],"it":[20],"struggles":[21],"with":[22,174],"perceiving":[23],"fine-grained":[24,70],"visual":[25,35,119,152],"details.":[26],"Generally,":[27],"to":[28,109,148],"enhance":[29],"representations,":[30],"take":[33],"CLIP's":[34],"features":[36],"as":[37,121],"conditions":[38,122],"for":[39,61,222],"reconstruction.":[40],"However,":[41],"the":[42,104,124,178,203,227],"underlying":[43],"principle":[44],"remains":[45],"underexplored.":[46],"In":[47],"this":[48],"work,":[49],"we":[50,83,142,155,187],"empirically":[51],"found":[52,92],"that":[53,93,115,133,157],"visually":[54],"perfect":[55],"generations":[56],"are":[57,231],"not":[58],"always":[59],"optimal":[60],"representation":[62],"enhancement.":[63],"essence":[65],"lies":[66],"effectively":[68],"extracting":[69],"knowledge":[71],"from":[72],"while":[75],"mitigating":[76],"irrelevant":[77],"information.":[78,138],"To":[79,139],"explore":[80,168],"critical":[81],"factors,":[82],"delve":[84],"into":[85,217],"three":[86],"aspects:":[87],"(1)":[88],"Conditioning":[89],"mechanisms:":[90],"We":[91,112,131,167],"even":[94],"a":[95,144],"small":[96],"number":[97],"of":[98,106,180],"local":[99],"tokens":[100,120],"can":[101,160,213],"drastically":[102],"reduce":[103],"difficulty":[105],"reconstruction,":[107],"leading":[108],"collapsed":[110],"training.":[111],"thus":[113],"conclude":[114],"utilizing":[116],"only":[117],"global":[118],"is":[123],"most":[125],"effective":[126,193],"strategy.":[127],"(2)":[128],"Denoising":[129],"configurations:":[130],"observed":[132],"end-to-end":[134],"training":[135,146],"introduces":[136],"extraneous":[137],"address":[140],"this,":[141],"propose":[143],"two-stage":[145],"strategy":[147],"prioritize":[149],"learning":[150],"useful":[151],"knowledge.":[153],"Additionally,":[154],"demonstrate":[156],"lightweight":[158],"denoisers":[159,173],"yield":[161],"remarkable":[162],"improvements.":[163],"(3)":[164],"Generation":[165],"paradigms:":[166],"both":[169],"continuous":[170],"discrete":[172],"desirable":[175],"outcomes,":[176],"validating":[177],"versatility":[179],"our":[181,184],"method.":[182],"Through":[183],"in-depth":[185],"explorations,":[186],"have":[188],"finally":[189],"arrived":[190],"at":[191],"an":[192],"method,":[194],"namely":[195],"GenHancer,":[196],"which":[197],"consistently":[198],"outperforms":[199],"prior":[200],"arts":[201],"on":[202,208],"MMVP-VLM":[204],"benchmark,":[205],"e.g.,":[206],"6.0%":[207],"OpenAICLIP.":[209],"enhanced":[211],"CLIP":[212],"be":[214],"further":[215],"plugged":[216],"multimodal":[218],"large":[219],"language":[220],"better":[223],"vision-centric":[224],"performance.":[225],"All":[226],"codes":[230],"made":[232],"publicly":[233],"available.":[234]},"counts_by_year":[],"updated_date":"2026-07-14T23:27:15.235271","created_date":"2025-10-10T00:00:00"}