{"id":"https://openalex.org/W7140216723","doi":"https://doi.org/10.48550/arxiv.2603.22027","title":"Tuning Real-World Image Restoration at Inference: A Test-Time Scaling Paradigm for Flow Matching Models","display_name":"Tuning Real-World Image Restoration at Inference: A Test-Time Scaling Paradigm for Flow Matching Models","publication_year":2026,"publication_date":"2026-03-23","ids":{"openalex":"https://openalex.org/W7140216723","doi":"https://doi.org/10.48550/arxiv.2603.22027"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2603.22027","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.22027","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,"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.2603.22027","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":null,"display_name":"Bai, Purui","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Bai, Purui","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Duan, Junxian","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Duan, Junxian","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Wang, Pin","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Pin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Hao, Jinhua","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Hao, Jinhua","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Sun, Ming","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Sun, Ming","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Zhou, Chao","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhou, Chao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":null,"display_name":"Huang, Huaibo","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Huang, Huaibo","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":0.578000009059906,"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":0.578000009059906,"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/T11105","display_name":"Advanced Image Processing Techniques","score":0.25040000677108765,"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/T11019","display_name":"Image Enhancement Techniques","score":0.028200000524520874,"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-restoration","display_name":"Image restoration","score":0.6776999831199646},{"id":"https://openalex.org/keywords/scaling","display_name":"Scaling","score":0.6064000129699707},{"id":"https://openalex.org/keywords/matching","display_name":"Matching (statistics)","score":0.49570000171661377},{"id":"https://openalex.org/keywords/image","display_name":"Image (mathematics)","score":0.46149998903274536},{"id":"https://openalex.org/keywords/transformer","display_name":"Transformer","score":0.3937000036239624},{"id":"https://openalex.org/keywords/sequence","display_name":"Sequence (biology)","score":0.3555000126361847},{"id":"https://openalex.org/keywords/encoding","display_name":"Encoding (memory)","score":0.35040000081062317}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6836000084877014},{"id":"https://openalex.org/C106430172","wikidata":"https://www.wikidata.org/wiki/Q6002272","display_name":"Image restoration","level":4,"score":0.6776999831199646},{"id":"https://openalex.org/C99844830","wikidata":"https://www.wikidata.org/wiki/Q102441924","display_name":"Scaling","level":2,"score":0.6064000129699707},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4999000132083893},{"id":"https://openalex.org/C165064840","wikidata":"https://www.wikidata.org/wiki/Q1321061","display_name":"Matching (statistics)","level":2,"score":0.49570000171661377},{"id":"https://openalex.org/C115961682","wikidata":"https://www.wikidata.org/wiki/Q860623","display_name":"Image (mathematics)","level":2,"score":0.46149998903274536},{"id":"https://openalex.org/C66322947","wikidata":"https://www.wikidata.org/wiki/Q11658","display_name":"Transformer","level":3,"score":0.3937000036239624},{"id":"https://openalex.org/C31972630","wikidata":"https://www.wikidata.org/wiki/Q844240","display_name":"Computer vision","level":1,"score":0.37940001487731934},{"id":"https://openalex.org/C2778112365","wikidata":"https://www.wikidata.org/wiki/Q3511065","display_name":"Sequence (biology)","level":2,"score":0.3555000126361847},{"id":"https://openalex.org/C125411270","wikidata":"https://www.wikidata.org/wiki/Q18653","display_name":"Encoding (memory)","level":2,"score":0.35040000081062317},{"id":"https://openalex.org/C163294075","wikidata":"https://www.wikidata.org/wiki/Q581861","display_name":"Noise reduction","level":2,"score":0.3488999903202057},{"id":"https://openalex.org/C155542232","wikidata":"https://www.wikidata.org/wiki/Q736111","display_name":"Optical flow","level":3,"score":0.3327000141143799},{"id":"https://openalex.org/C38349280","wikidata":"https://www.wikidata.org/wiki/Q1434290","display_name":"Flow (mathematics)","level":2,"score":0.32899999618530273},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.3278999924659729},{"id":"https://openalex.org/C69744172","wikidata":"https://www.wikidata.org/wiki/Q860822","display_name":"Image fusion","level":3,"score":0.32280001044273376},{"id":"https://openalex.org/C9417928","wikidata":"https://www.wikidata.org/wiki/Q1070689","display_name":"Image processing","level":3,"score":0.32120001316070557},{"id":"https://openalex.org/C158525013","wikidata":"https://www.wikidata.org/wiki/Q2593739","display_name":"Fusion","level":2,"score":0.2946000099182129},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.2531000077724457}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2603.22027","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.22027","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,"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.2603.22027","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.22027","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,"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":[{"id":"https://metadata.un.org/sdg/11","display_name":"Sustainable cities and communities","score":0.7059759497642517}],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Although":[0],"diffusion-based":[1],"real-world":[2],"image":[3,33,97],"restoration":[4,34,57],"(Real-IR)":[5],"has":[6],"achieved":[7],"remarkable":[8],"progress,":[9],"efficiently":[10],"leveraging":[11],"ultra-large-scale":[12],"pre-trained":[13,168],"text-to-image":[14],"(T2I)":[15],"models":[16],"and":[17,160],"fully":[18,61],"exploiting":[19],"their":[20],"potential":[21],"remain":[22],"significant":[23,117],"challenges.":[24],"To":[25],"address":[26],"this":[27,101],"issue,":[28],"we":[29,88],"propose":[30],"ResFlow-Tuner,":[31],"an":[32],"framework":[35],"based":[36],"on":[37],"the":[38,63,66,82,105,142,146],"state-of-the-art":[39,131],"flow":[40,147],"matching":[41,148],"model,":[42],"FLUX.1-dev,":[43],"which":[44],"integrates":[45],"unified":[46,78],"multi-modal":[47,74],"fusion":[48],"(UMMF)":[49],"with":[50,120],"test-time":[51,92],"scaling":[52,93,163],"(TTS)":[53],"to":[54],"achieve":[55],"unprecedented":[56],"performance.":[58],"Our":[59],"approach":[60],"leverages":[62],"advantages":[64],"of":[65,84,145],"Multi-Modal":[67],"Diffusion":[68],"Transformer":[69],"(MM-DiT)":[70],"architecture":[71],"by":[72],"encoding":[73],"conditions":[75],"into":[76],"a":[77,90,111,158],"sequence":[79],"that":[80,127],"guides":[81],"synthesis":[83],"high-quality":[85],"images.":[86],"Furthermore,":[87],"introduce":[89],"training-free":[91],"paradigm":[94,164],"tailored":[95],"for":[96,166],"restoration.":[98],"During":[99],"inference,":[100],"technique":[102],"dynamically":[103],"steers":[104],"denoising":[106],"direction":[107],"through":[108],"feedback":[109],"from":[110],"reward":[112],"model":[113,149],"(RM),":[114],"thereby":[115],"achieving":[116],"performance":[118,132],"gains":[119],"controllable":[121],"computational":[122],"overhead.":[123],"Extensive":[124],"experiments":[125],"demonstrate":[126],"our":[128],"method":[129],"achieves":[130],"across":[133],"multiple":[134],"standard":[135],"benchmarks.":[136],"This":[137],"work":[138],"not":[139],"only":[140],"validates":[141],"powerful":[143],"capabilities":[144],"in":[150],"low-level":[151],"vision":[152],"tasks":[153],"but,":[154],"more":[155],"importantly,":[156],"proposes":[157],"novel":[159],"efficient":[161],"inference-time":[162],"suitable":[165],"large":[167],"models.":[169]},"counts_by_year":[],"updated_date":"2026-07-28T07:46:37.118299","created_date":"2026-03-25T00:00:00"}