{"id":"https://openalex.org/W7166869986","doi":"https://doi.org/10.48550/arxiv.2606.31603","title":"Preserve the Hard, Regenerate the Rest: Uncertainty-Guided Synthetic Training Data Augmentation with Diffusion Models","display_name":"Preserve the Hard, Regenerate the Rest: Uncertainty-Guided Synthetic Training Data Augmentation with Diffusion Models","publication_year":2026,"publication_date":"2026-06-30","ids":{"openalex":"https://openalex.org/W7166869986","doi":"https://doi.org/10.48550/arxiv.2606.31603"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2606.31603","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.31603","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.2606.31603","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5119990135","display_name":"Nikolai R\u00f6hrich","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"R\u00f6hrich, Nikolai","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5120734578","display_name":"Julian Glei\u00dfner","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Glei\u00dfner, Julian","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5060682816","display_name":"Adam Ibrahim","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ibrahim, Ahmed H. A.","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5043580137","display_name":"Silvan Mertes","orcid":"https://orcid.org/0000-0001-5230-5218"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Mertes, Silvan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5139749579","display_name":"Tobias Huber","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Huber, Tobias","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/T10036","display_name":"Advanced Neural Network Applications","score":0.9172000288963318,"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/T10036","display_name":"Advanced Neural Network Applications","score":0.9172000288963318,"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/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.03420000150799751,"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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.007600000128149986,"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/segmentation","display_name":"Segmentation","score":0.7080000042915344},{"id":"https://openalex.org/keywords/context","display_name":"Context (archaeology)","score":0.6668000221252441},{"id":"https://openalex.org/keywords/synthetic-data","display_name":"Synthetic data","score":0.6025999784469604},{"id":"https://openalex.org/keywords/heuristics","display_name":"Heuristics","score":0.6008999943733215},{"id":"https://openalex.org/keywords/training-set","display_name":"Training set","score":0.4269999861717224},{"id":"https://openalex.org/keywords/code","display_name":"Code (set theory)","score":0.3887999951839447}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.748199999332428},{"id":"https://openalex.org/C89600930","wikidata":"https://www.wikidata.org/wiki/Q1423946","display_name":"Segmentation","level":2,"score":0.7080000042915344},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.667900025844574},{"id":"https://openalex.org/C2779343474","wikidata":"https://www.wikidata.org/wiki/Q3109175","display_name":"Context (archaeology)","level":2,"score":0.6668000221252441},{"id":"https://openalex.org/C160920958","wikidata":"https://www.wikidata.org/wiki/Q7662746","display_name":"Synthetic data","level":2,"score":0.6025999784469604},{"id":"https://openalex.org/C127705205","wikidata":"https://www.wikidata.org/wiki/Q5748245","display_name":"Heuristics","level":2,"score":0.6008999943733215},{"id":"https://openalex.org/C51632099","wikidata":"https://www.wikidata.org/wiki/Q3985153","display_name":"Training set","level":2,"score":0.4269999861717224},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.4262000024318695},{"id":"https://openalex.org/C2776760102","wikidata":"https://www.wikidata.org/wiki/Q5139990","display_name":"Code (set theory)","level":3,"score":0.3887999951839447},{"id":"https://openalex.org/C124504099","wikidata":"https://www.wikidata.org/wiki/Q56933","display_name":"Image segmentation","level":3,"score":0.3797000050544739},{"id":"https://openalex.org/C31972630","wikidata":"https://www.wikidata.org/wiki/Q844240","display_name":"Computer vision","level":1,"score":0.36649999022483826},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.3379000127315521},{"id":"https://openalex.org/C99498987","wikidata":"https://www.wikidata.org/wiki/Q2210247","display_name":"Noise (video)","level":3,"score":0.3086000084877014},{"id":"https://openalex.org/C160633673","wikidata":"https://www.wikidata.org/wiki/Q355198","display_name":"Pixel","level":2,"score":0.2937999963760376},{"id":"https://openalex.org/C12725497","wikidata":"https://www.wikidata.org/wiki/Q810247","display_name":"Baseline (sea)","level":2,"score":0.29190000891685486},{"id":"https://openalex.org/C200873422","wikidata":"https://www.wikidata.org/wiki/Q5448821","display_name":"Filling-in","level":2,"score":0.25029999017715454}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2606.31603","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.31603","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.2606.31603","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.31603","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/10","display_name":"Reduced inequalities","score":0.6128365993499756}],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Semantic":[0],"segmentation":[1,198],"models":[2],"struggle":[3],"with":[4,165,203],"data":[5,36],"sparsity":[6],"and":[7,41,90,114,163,171],"rare":[8,170],"or":[9,16,21,54,65,178],"visually":[10],"diverse":[11],"regions,":[12],"e.g.,":[13],"dense":[14],"regions":[15,113,148],"small":[17],"objects":[18,64],"in":[19,152],"aerial":[20,181],"autonomous":[22],"mobility":[23],"data.":[24],"While":[25],"synthetic":[26,81,96,127],"augmentation":[27,83,190],"is":[28,191],"an":[29,79],"appealing":[30],"solution,":[31],"directly":[32],"generating":[33],"new":[34],"labeled":[35],"risks":[37],"misalignment":[38],"of":[39],"labels":[40],"generated":[42],"pixels.":[43,73],"Existing":[44],"solutions":[45],"to":[46,196],"this":[47,126],"problem":[48],"often":[49],"rely":[50],"on":[51,71,125,144,160,169,200],"external":[52,100],"models,":[53],"employ":[55],"coarse":[56],"heuristics":[57],"such":[58,174],"as":[59,175],"indiscriminately":[60],"augmenting":[61],"all":[62],"foreground":[63],"entire":[66],"backgrounds,":[67],"which":[68],"wastes":[69],"capacity":[70],"uninformative":[72],"To":[74],"address":[75],"this,":[76],"we":[77,109,129],"propose":[78],"uncertainty-guided":[80,188],"context":[82,189],"strategy":[84],"that":[85,187],"strictly":[86],"preserves":[87],"label":[88],"validity":[89],"efficiently":[91],"maximizes":[92],"pixel":[93],"informativeness":[94],"per":[95],"sample":[97],"-":[98],"no":[99],"guardrails":[101],"required.":[102],"Using":[103],"a":[104,192],"baseline":[105],"segmenter's":[106],"predictive":[107],"entropy,":[108],"identify":[110],"uncertain":[111,147],"semantic":[112],"inpaint":[115],"only":[116,133],"the":[117,123,131,135,145,166,180],"complementary":[118],"visual":[119],"context.":[120],"When":[121],"fine-tuning":[122],"segmenter":[124],"data,":[128],"compute":[130],"loss":[132],"over":[134],"original":[136],"pixels,":[137],"excluding":[138],"inpainted":[139],"regions.":[140],"This":[141],"focuses":[142],"learning":[143],"unmodified,":[146],"while":[149],"presenting":[150],"them":[151],"novel":[153],"contexts.":[154],"We":[155],"demonstrate":[156,186],"substantial":[157],"mIoU":[158],"gains":[159,168],"Cityscapes,":[161],"UAVID,":[162],"BDD100K":[164],"largest":[167],"difficult":[172],"classes":[173],"buses,":[176],"trains,":[177],"(from":[179],"perspective)":[182],"cars.":[183],"Our":[184],"results":[185],"highly":[193],"effective":[194],"lever":[195],"improve":[197],"performance":[199],"complex":[201],"datasets,":[202],"code":[204],"provided":[205],"at":[206],"https://github.com/XITASO/Preserve-the-Hard-Regenerate-the-Rest.":[207]},"counts_by_year":[],"updated_date":"2026-07-28T07:46:37.118299","created_date":"2026-07-02T00:00:00"}