{"id":"https://openalex.org/W7162408507","doi":"https://doi.org/10.48550/arxiv.2605.25046","title":"TinyFormer: Preserving Tiny Objects in YOLO-DETR Hybrid Real-time Detectors","display_name":"TinyFormer: Preserving Tiny Objects in YOLO-DETR Hybrid Real-time Detectors","publication_year":2026,"publication_date":"2026-05-24","ids":{"openalex":"https://openalex.org/W7162408507","doi":"https://doi.org/10.48550/arxiv.2605.25046"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2605.25046","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.25046","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.2605.25046","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5137039644","display_name":"Jun-Wei Hsieh","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Hsieh, Jun-Wei","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5012773990","display_name":"Meng-Yu Kao","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Kao, Meng-Yu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5080252492","display_name":"Ghufron Wahyu Kurniawan","orcid":"https://orcid.org/0000-0002-8178-2926"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Kurniawan, Ghufron Wahyu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5063443405","display_name":"Kuan\u2013Chuan Peng","orcid":"https://orcid.org/0000-0002-2682-9912"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Peng, Kuan-Chuan","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":0.5806000232696533,"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":0.5806000232696533,"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":0.26600000262260437,"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":0.01209999993443489,"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/security-token","display_name":"Security token","score":0.7186999917030334},{"id":"https://openalex.org/keywords/detector","display_name":"Detector","score":0.6517000198364258},{"id":"https://openalex.org/keywords/feature","display_name":"Feature (linguistics)","score":0.5404999852180481},{"id":"https://openalex.org/keywords/grid","display_name":"Grid","score":0.5220000147819519},{"id":"https://openalex.org/keywords/set","display_name":"Set (abstract data type)","score":0.4771000146865845},{"id":"https://openalex.org/keywords/pyramid","display_name":"Pyramid (geometry)","score":0.39419999718666077},{"id":"https://openalex.org/keywords/code","display_name":"Code (set theory)","score":0.37549999356269836}],"concepts":[{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.7186999917030334},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7181000113487244},{"id":"https://openalex.org/C94915269","wikidata":"https://www.wikidata.org/wiki/Q1834857","display_name":"Detector","level":2,"score":0.6517000198364258},{"id":"https://openalex.org/C2776401178","wikidata":"https://www.wikidata.org/wiki/Q12050496","display_name":"Feature (linguistics)","level":2,"score":0.5404999852180481},{"id":"https://openalex.org/C187691185","wikidata":"https://www.wikidata.org/wiki/Q2020720","display_name":"Grid","level":2,"score":0.5220000147819519},{"id":"https://openalex.org/C177264268","wikidata":"https://www.wikidata.org/wiki/Q1514741","display_name":"Set (abstract data type)","level":2,"score":0.4771000146865845},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4586000144481659},{"id":"https://openalex.org/C142575187","wikidata":"https://www.wikidata.org/wiki/Q3358290","display_name":"Pyramid (geometry)","level":2,"score":0.39419999718666077},{"id":"https://openalex.org/C31972630","wikidata":"https://www.wikidata.org/wiki/Q844240","display_name":"Computer vision","level":1,"score":0.376800000667572},{"id":"https://openalex.org/C2776760102","wikidata":"https://www.wikidata.org/wiki/Q5139990","display_name":"Code (set theory)","level":3,"score":0.37549999356269836},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.3747999966144562},{"id":"https://openalex.org/C66322947","wikidata":"https://www.wikidata.org/wiki/Q11658","display_name":"Transformer","level":3,"score":0.35899999737739563},{"id":"https://openalex.org/C52622490","wikidata":"https://www.wikidata.org/wiki/Q1026626","display_name":"Feature extraction","level":2,"score":0.35040000081062317},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.31929999589920044},{"id":"https://openalex.org/C113775141","wikidata":"https://www.wikidata.org/wiki/Q428691","display_name":"Computer engineering","level":1,"score":0.30000001192092896},{"id":"https://openalex.org/C184337299","wikidata":"https://www.wikidata.org/wiki/Q1437428","display_name":"Semantics (computer science)","level":2,"score":0.2833000123500824},{"id":"https://openalex.org/C2776151529","wikidata":"https://www.wikidata.org/wiki/Q3045304","display_name":"Object detection","level":3,"score":0.27219998836517334},{"id":"https://openalex.org/C152124472","wikidata":"https://www.wikidata.org/wiki/Q1204361","display_name":"Redundancy (engineering)","level":2,"score":0.2694000005722046},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.25690001249313354}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2605.25046","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.25046","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.2605.25046","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.25046","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":[{"score":0.4751894176006317,"display_name":"Decent work and economic growth","id":"https://metadata.un.org/sdg/8"}],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"YOLO-series":[0,169],"and":[1,27,56,82,141,171,192,210,217,230],"DETR-based":[2,32],"detectors":[3,170,213],"struggle":[4],"with":[5,214],"tiny-object":[6,149,241],"detection.":[7,90,242],"YOLO-style":[8,84,227],"models":[9,33],"benefit":[10],"from":[11,102,138],"efficient":[12],"dense":[13,226],"prediction,":[14,39,81,233],"but":[15],"their":[16],"large-stride":[17],"backbones":[18],"may":[19],"suppress":[20],"tiny":[21,48],"instances":[22],"in":[23],"deep":[24],"feature":[25,107,228],"maps":[26],"make":[28],"grid":[29],"assignment":[30],"ambiguous.":[31],"remove":[34],"hand-crafted":[35],"post-processing":[36],"through":[37],"set":[38,80,232],"yet":[40],"they":[41],"reason":[42],"over":[43],"coarse":[44,132],"token":[45,146],"grids,":[46],"where":[47],"objects":[49],"occupy":[50],"only":[51],"a":[52,69,83,93,119,194,235],"few":[53],"weak":[54],"tokens":[55],"are":[57],"easily":[58],"overlooked":[59],"during":[60,113],"matching.":[61],"To":[62],"address":[63],"these":[64],"limitations,":[65],"we":[66],"propose":[67],"TinyFormer,":[68],"unified":[70],"YOLO--DETR":[71],"hybrid":[72],"real-time":[73,240],"detector":[74],"that":[75,164,223],"combines":[76],"ViT":[77],"representations,":[78],"NMS-free":[79],"pyramid":[85],"neck":[86],"for":[87,126,239],"accurate":[88],"small-object":[89],"TinyFormer":[91,165,224],"introduces":[92],"Parallel":[94],"Bi-fusion":[95],"Module":[96],"(PBM),":[97],"which":[98],"builds":[99],"high-resolution":[100,136],"shortcuts":[101],"shallow":[103],"stages":[104,140],"to":[105,124,190],"the":[106,127,153,172,187],"pyramid,":[108],"preserving":[109],"fine":[110],"spatial":[111,128],"details":[112],"multi-scale":[114],"fusion.":[115],"We":[116],"further":[117],"design":[118],"Spatial":[120],"Semantic":[121],"Adapter":[122],"(SSA)":[123],"compensate":[125],"loss":[129],"caused":[130],"by":[131],"tokenization.":[133],"SSA":[134],"extracts":[135],"cues":[137],"early":[139],"injects":[142],"them":[143],"into":[144],"transformer":[145],"embeddings,":[147],"improving":[148],"localization":[150],"without":[151,181],"sacrificing":[152],"global":[154],"modeling":[155],"ability":[156],"of":[157],"DETR.":[158],"Experiments":[159],"on":[160,198],"MS":[161],"COCO":[162],"show":[163],"consistently":[166],"outperforms":[167],"recent":[168],"strong":[173,236],"DEIMv2":[174],"baseline.":[175],"TinyFormer-X":[176],"achieves":[177],"58.4%":[178],"AP":[179,189,196],"even":[180],"PBM,":[182],"while":[183],"adding":[184],"PBM":[185],"improves":[186],"overall":[188],"58.5%":[191],"brings":[193],"1.6%":[195],"gain":[197],"small":[199],"objects.":[200],"With":[201],"Objects365":[202],"pre-training,":[203],"TinyFormer-X-PBM":[204],"reaches":[205],"60.2%":[206],"AP,":[207],"surpassing":[208],"RF-DETR":[209],"other":[211],"Objects365-pretrained":[212],"fewer":[215],"parameters":[216],"lower":[218],"computation.":[219],"These":[220],"results":[221],"demonstrate":[222],"bridges":[225],"fusion":[229],"DETR-style":[231],"providing":[234],"accuracy-efficiency":[237],"trade-off":[238],"Code":[243],"is":[244],"available":[245],"at":[246],"https://github.com/mmpmmpmmpjosh/TinyFormer.":[247]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-05-27T00:00:00"}