{"id":"https://openalex.org/W4389667009","doi":"https://doi.org/10.1109/iros55552.2023.10341422","title":"Early or Late Fusion Matters: Efficient RGB-D Fusion in Vision Transformers for 3D Object Recognition","display_name":"Early or Late Fusion Matters: Efficient RGB-D Fusion in Vision Transformers for 3D Object Recognition","publication_year":2023,"publication_date":"2023-10-01","ids":{"openalex":"https://openalex.org/W4389667009","doi":"https://doi.org/10.1109/iros55552.2023.10341422"},"language":"en","primary_location":{"id":"doi:10.1109/iros55552.2023.10341422","is_oa":false,"landing_page_url":"https://doi.org/10.1109/iros55552.2023.10341422","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2023 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)","raw_type":"proceedings-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://research.rug.nl/en/publications/80edaa9a-0832-43a9-acde-42beffbb787a","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5004690448","display_name":"Georgios Tziafas","orcid":null},"institutions":[{"id":"https://openalex.org/I169381384","display_name":"University of Groningen","ror":"https://ror.org/012p63287","country_code":"NL","type":"education","lineage":["https://openalex.org/I169381384"]}],"countries":["NL"],"is_corresponding":true,"raw_author_name":"Georgios Tziafas","raw_affiliation_strings":["University of Groningen,Department of Artificial Intelligence,The Netherlands","Department of Artificial Intelligence, University of Groningen, The Netherlands"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Groningen,Department of Artificial Intelligence,The Netherlands","institution_ids":["https://openalex.org/I169381384"]},{"raw_affiliation_string":"Department of Artificial Intelligence, University of Groningen, The Netherlands","institution_ids":["https://openalex.org/I169381384"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5008655688","display_name":"Hamidreza Kasaei","orcid":null},"institutions":[{"id":"https://openalex.org/I169381384","display_name":"University of Groningen","ror":"https://ror.org/012p63287","country_code":"NL","type":"education","lineage":["https://openalex.org/I169381384"]}],"countries":["NL"],"is_corresponding":false,"raw_author_name":"Hamidreza Kasaei","raw_affiliation_strings":["University of Groningen,Department of Artificial Intelligence,The Netherlands","Department of Artificial Intelligence, University of Groningen, The Netherlands"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Groningen,Department of Artificial Intelligence,The Netherlands","institution_ids":["https://openalex.org/I169381384"]},{"raw_affiliation_string":"Department of Artificial Intelligence, University of Groningen, The Netherlands","institution_ids":["https://openalex.org/I169381384"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":2,"corresponding_author_ids":["https://openalex.org/A5004690448"],"corresponding_institution_ids":["https://openalex.org/I169381384"],"apc_list":null,"apc_paid":null,"fwci":2.2268,"has_fulltext":false,"cited_by_count":19,"citation_normalized_percentile":{"value":0.90196206,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":98,"max":99},"biblio":{"volume":null,"issue":null,"first_page":"9558","last_page":"9565"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.9998000264167786,"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":0.9998000264167786,"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":0.9994999766349792,"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/T10191","display_name":"Robotics and Sensor-Based Localization","score":0.9969000220298767,"subfield":{"id":"https://openalex.org/subfields/2202","display_name":"Aerospace Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/rgb-color-model","display_name":"RGB color model","score":0.7793095111846924},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.7714791893959045},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.7388757467269897},{"id":"https://openalex.org/keywords/computer-vision","display_name":"Computer vision","score":0.6082580089569092},{"id":"https://openalex.org/keywords/benchmark","display_name":"Benchmark (surveying)","score":0.5338270664215088},{"id":"https://openalex.org/keywords/fusion-mechanism","display_name":"Fusion mechanism","score":0.5136327147483826},{"id":"https://openalex.org/keywords/transformer","display_name":"Transformer","score":0.4414350390434265},{"id":"https://openalex.org/keywords/fusion","display_name":"Fusion","score":0.35148561000823975},{"id":"https://openalex.org/keywords/engineering","display_name":"Engineering","score":0.11932951211929321}],"concepts":[{"id":"https://openalex.org/C82990744","wikidata":"https://www.wikidata.org/wiki/Q166194","display_name":"RGB color model","level":2,"score":0.7793095111846924},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.7714791893959045},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7388757467269897},{"id":"https://openalex.org/C31972630","wikidata":"https://www.wikidata.org/wiki/Q844240","display_name":"Computer vision","level":1,"score":0.6082580089569092},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.5338270664215088},{"id":"https://openalex.org/C173414695","wikidata":"https://www.wikidata.org/wiki/Q5510276","display_name":"Fusion mechanism","level":4,"score":0.5136327147483826},{"id":"https://openalex.org/C66322947","wikidata":"https://www.wikidata.org/wiki/Q11658","display_name":"Transformer","level":3,"score":0.4414350390434265},{"id":"https://openalex.org/C158525013","wikidata":"https://www.wikidata.org/wiki/Q2593739","display_name":"Fusion","level":2,"score":0.35148561000823975},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.11932951211929321},{"id":"https://openalex.org/C103038307","wikidata":"https://www.wikidata.org/wiki/Q6556360","display_name":"Lipid bilayer fusion","level":3,"score":0.0},{"id":"https://openalex.org/C119599485","wikidata":"https://www.wikidata.org/wiki/Q43035","display_name":"Electrical engineering","level":1,"score":0.0},{"id":"https://openalex.org/C205649164","wikidata":"https://www.wikidata.org/wiki/Q1071","display_name":"Geography","level":0,"score":0.0},{"id":"https://openalex.org/C165801399","wikidata":"https://www.wikidata.org/wiki/Q25428","display_name":"Voltage","level":2,"score":0.0},{"id":"https://openalex.org/C138885662","wikidata":"https://www.wikidata.org/wiki/Q5891","display_name":"Philosophy","level":0,"score":0.0},{"id":"https://openalex.org/C41895202","wikidata":"https://www.wikidata.org/wiki/Q8162","display_name":"Linguistics","level":1,"score":0.0},{"id":"https://openalex.org/C13280743","wikidata":"https://www.wikidata.org/wiki/Q131089","display_name":"Geodesy","level":1,"score":0.0}],"mesh":[],"locations_count":3,"locations":[{"id":"doi:10.1109/iros55552.2023.10341422","is_oa":false,"landing_page_url":"https://doi.org/10.1109/iros55552.2023.10341422","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2023 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)","raw_type":"proceedings-article"},{"id":"pmh:oai:pure.rug.nl:openaire_cris_publications/80edaa9a-0832-43a9-acde-42beffbb787a","is_oa":true,"landing_page_url":"https://research.rug.nl/en/publications/80edaa9a-0832-43a9-acde-42beffbb787a","pdf_url":null,"source":{"id":"https://openalex.org/S4306400420","display_name":"University of Groningen research database (University of Groningen / Centre for Information Technology)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I169381384","host_organization_name":"University of Groningen","host_organization_lineage":["https://openalex.org/I169381384"],"host_organization_lineage_names":[],"type":"repository"},"license":"other-oa","license_id":"https://openalex.org/licenses/other-oa","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Tziafas, G & Kasaei, H 2023, Early or Late Fusion Matters : Efficient RGB-D Fusion in Vision Transformers for 3D Object Recognition. in 2023 IEEE/RSJ International Conference on Intelligent Robots and Systems, IROS 2023. IEEE International Conference on Intelligent Robots and Systems, Institute of Electrical and Electronics Engineers Inc., pp. 9558-9565, 2023 IEEE/RSJ International Conference on Intelligent Robots and Systems, IROS 2023, Detroit, United States, 01/10/2023. https://doi.org/10.1109/IROS55552.2023.10341422","raw_type":"info:eu-repo/semantics/publishedVersion"},{"id":"pmh:oai:pure.rug.nl:openaire/80edaa9a-0832-43a9-acde-42beffbb787a","is_oa":true,"landing_page_url":"https://hdl.handle.net/11370/80edaa9a-0832-43a9-acde-42beffbb787a","pdf_url":null,"source":{"id":"https://openalex.org/S4306400420","display_name":"University of Groningen research database (University of Groningen / Centre for Information Technology)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I169381384","host_organization_name":"University of Groningen","host_organization_lineage":["https://openalex.org/I169381384"],"host_organization_lineage_names":[],"type":"repository"},"license":"other-oa","license_id":"https://openalex.org/licenses/other-oa","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"Tziafas, G & Kasaei, H 2023, Early or Late Fusion Matters : Efficient RGB-D Fusion in Vision Transformers for 3D Object Recognition. in 2023 IEEE/RSJ International Conference on Intelligent Robots and Systems, IROS 2023. IEEE International Conference on Intelligent Robots and Systems, Institute of Electrical and Electronics Engineers Inc., pp. 9558-9565, 2023 IEEE/RSJ International Conference on Intelligent Robots and Systems, IROS 2023, Detroit, United States, 01/10/2023. https://doi.org/10.1109/IROS55552.2023.10341422","raw_type":"info:eu-repo/semantics/publishedVersion"}],"best_oa_location":{"id":"pmh:oai:pure.rug.nl:openaire_cris_publications/80edaa9a-0832-43a9-acde-42beffbb787a","is_oa":true,"landing_page_url":"https://research.rug.nl/en/publications/80edaa9a-0832-43a9-acde-42beffbb787a","pdf_url":null,"source":{"id":"https://openalex.org/S4306400420","display_name":"University of Groningen research database (University of Groningen / Centre for Information Technology)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I169381384","host_organization_name":"University of Groningen","host_organization_lineage":["https://openalex.org/I169381384"],"host_organization_lineage_names":[],"type":"repository"},"license":"other-oa","license_id":"https://openalex.org/licenses/other-oa","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Tziafas, G & Kasaei, H 2023, Early or Late Fusion Matters : Efficient RGB-D Fusion in Vision Transformers for 3D Object Recognition. in 2023 IEEE/RSJ International Conference on Intelligent Robots and Systems, IROS 2023. IEEE International Conference on Intelligent Robots and Systems, Institute of Electrical and Electronics Engineers Inc., pp. 9558-9565, 2023 IEEE/RSJ International Conference on Intelligent Robots and Systems, IROS 2023, Detroit, United States, 01/10/2023. https://doi.org/10.1109/IROS55552.2023.10341422","raw_type":"info:eu-repo/semantics/publishedVersion"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":52,"referenced_works":["https://openalex.org/W1505952289","https://openalex.org/W1591677984","https://openalex.org/W1884730573","https://openalex.org/W1922904362","https://openalex.org/W2074142320","https://openalex.org/W2108598243","https://openalex.org/W2156222070","https://openalex.org/W2165599843","https://openalex.org/W2176371886","https://openalex.org/W2207044458","https://openalex.org/W2551737796","https://openalex.org/W2752585553","https://openalex.org/W2774550181","https://openalex.org/W2788515346","https://openalex.org/W2789835518","https://openalex.org/W2860627718","https://openalex.org/W2902625698","https://openalex.org/W2911788474","https://openalex.org/W2920456178","https://openalex.org/W2962843773","https://openalex.org/W2963011558","https://openalex.org/W2963032410","https://openalex.org/W2963901718","https://openalex.org/W2963956866","https://openalex.org/W2964189064","https://openalex.org/W2967153639","https://openalex.org/W3011972190","https://openalex.org/W3019632826","https://openalex.org/W3040130969","https://openalex.org/W3094488141","https://openalex.org/W3125116114","https://openalex.org/W3138516171","https://openalex.org/W3157898411","https://openalex.org/W3170863103","https://openalex.org/W3170874841","https://openalex.org/W4206706211","https://openalex.org/W4281383322","https://openalex.org/W4287203292","https://openalex.org/W4292794834","https://openalex.org/W4312372834","https://openalex.org/W4312633518","https://openalex.org/W4386071576","https://openalex.org/W4386179772","https://openalex.org/W6684489972","https://openalex.org/W6730146409","https://openalex.org/W6730320844","https://openalex.org/W6757384668","https://openalex.org/W6757817989","https://openalex.org/W6788135285","https://openalex.org/W6794559225","https://openalex.org/W6796761347","https://openalex.org/W6810856158"],"related_works":["https://openalex.org/W2988498288","https://openalex.org/W1991764124","https://openalex.org/W2786308501","https://openalex.org/W2012247345","https://openalex.org/W4318719019","https://openalex.org/W2145243726","https://openalex.org/W2351649588","https://openalex.org/W4378188910","https://openalex.org/W1998943912","https://openalex.org/W2463550667"],"abstract_inverted_index":{"The":[0],"Vision":[1],"Transformer":[2],"(ViT)":[3],"architecture":[4],"has":[5],"established":[6],"its":[7],"place":[8],"in":[9,25,35,73,96,130,155,164,193,199,217,225,231,265,272],"computer":[10],"vision":[11,37],"literature,":[12],"however,":[13],"training":[14],"ViTs":[15,72,110,186],"for":[16,69,76,142],"RGB-D":[17,74,158,168],"object":[18,78],"recognition":[19],"remains":[20],"an":[21,226,266],"understudied":[22],"topic,":[23],"viewed":[24],"recent":[26],"literature":[27],"only":[28],"through":[29],"the":[30,47,90,99,106,116,120,144,150,156,205,232],"lens":[31],"of":[32,49,109,132,207,244],"multi-task":[33],"pretraining":[34,51,121],"multiple":[36,50],"modalities.":[38],"Such":[39],"approaches":[40],"are":[41],"often":[42],"computationally":[43],"intensive,":[44],"relying":[45],"on":[46,81],"scale":[48],"datasets":[52],"to":[53,93,104,111,189],"align":[54],"RGB":[55,83,145,166],"with":[56,251,275],"3D":[57,77],"information.":[58],"In":[59],"this":[60,200],"work,":[61],"we":[62,247],"propose":[63],"a":[64,218,242,252,262,276],"simple":[65],"yet":[66],"strong":[67],"recipe":[68],"transferring":[70],"pretrained":[71],"domains":[75],"recognition,":[79],"focusing":[80],"fusing":[82],"and":[84,118,135,138,146,255,274],"depth":[85,126,147],"representations":[86],"encoded":[87],"jointly":[88],"by":[89,241],"ViT.":[91],"Compared":[92],"previous":[94,239],"works":[95,240],"multimodal":[97,210],"Transformers,":[98],"key":[100],"challenge":[101],"here":[102],"is":[103,128],"use":[105],"attested":[107],"flexibility":[108],"capture":[112],"cross-modal":[113],"interactions":[114],"at":[115],"downstream":[117],"not":[119],"stage.":[122],"We":[123,202],"explore":[124],"which":[125],"representation":[127],"better":[129,174],"terms":[131],"resulting":[133],"accuracy":[134,192],"compare":[136],"early":[137,179],"late":[139,170],"fusion":[140,171,185,211],"techniques":[141,172],"aligning":[143],"modalities":[148],"within":[149],"ViT":[151],"architecture.":[152],"Experimental":[153],"results":[154,198],"Washington":[157],"Objects":[159],"dataset":[160],"(ROD)":[161],"demonstrate":[162,256],"that":[163],"such":[165],"\u2192":[167],"scenarios,":[169],"work":[173],"than":[175],"most":[176],"popularly":[177],"employed":[178],"fusion.":[180],"With":[181],"our":[182,209,236,249],"transfer":[183],"baseline,":[184],"score":[187],"up":[188],"95.4%":[190],"top-1":[191],"ROD,":[194],"achieving":[195],"new":[196],"state-of-the-art":[197],"benchmark.":[201],"further":[203],"show":[204],"benefits":[206],"using":[208],"baseline":[212],"over":[213],"unimodal":[214],"feature":[215],"extractors":[216],"synthetic-to-real":[219],"visual":[220],"adaptation":[221],"as":[222,224,261],"well":[223],"open-ended":[227],"lifelong":[228],"learning":[229,269],"scenario":[230],"ROD":[233],"benchmark,":[234],"where":[235],"model":[237],"outperforms":[238],"margin":[243],">8%.":[245],"Finally,":[246],"integrate":[248],"method":[250],"robot":[253,268],"framework":[254],"how":[257],"it":[258],"can":[259],"serve":[260],"perception":[263],"utility":[264],"interactive":[267],"scenario,":[270],"both":[271],"simulation":[273],"real":[277],"robot.":[278]},"counts_by_year":[{"year":2026,"cited_by_count":2},{"year":2025,"cited_by_count":9},{"year":2024,"cited_by_count":8}],"updated_date":"2026-05-03T08:25:01.440150","created_date":"2025-10-10T00:00:00"}