{"id":"https://openalex.org/W7138011493","doi":"https://doi.org/10.1609/aaai.v40i13.38032","title":"Efficient Segmentation with Multimodal Large Language Model via Token Routing","display_name":"Efficient Segmentation with Multimodal Large Language Model via Token Routing","publication_year":2026,"publication_date":"2026-03-14","ids":{"openalex":"https://openalex.org/W7138011493","doi":"https://doi.org/10.1609/aaai.v40i13.38032"},"language":"en","primary_location":{"id":"doi:10.1609/aaai.v40i13.38032","is_oa":true,"landing_page_url":"https://doi.org/10.1609/aaai.v40i13.38032","pdf_url":"https://ojs.aaai.org/index.php/AAAI/article/download/38032/41994","source":{"id":"https://openalex.org/S4210191458","display_name":"Proceedings of the AAAI Conference on Artificial Intelligence","issn_l":"2159-5399","issn":["2159-5399","2374-3468"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/P4310320058","host_organization_name":"Association for the Advancement of Artificial Intelligence","host_organization_lineage":["https://openalex.org/P4310320058"],"host_organization_lineage_names":["Association for the Advancement of Artificial Intelligence"],"type":"conference"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the AAAI Conference on Artificial Intelligence","raw_type":"journal-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"diamond","oa_url":"https://ojs.aaai.org/index.php/AAAI/article/download/38032/41994","any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5108858621","display_name":"Changsong Wen","orcid":null},"institutions":[{"id":"https://openalex.org/I183067930","display_name":"Shanghai Jiao Tong University","ror":"https://ror.org/0220qvk04","country_code":"CN","type":"education","lineage":["https://openalex.org/I183067930"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Changsong Wen","raw_affiliation_strings":["Shanghai Jiao Tong University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Shanghai Jiao Tong University","institution_ids":["https://openalex.org/I183067930"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5016326528","display_name":"Zelin Peng","orcid":"https://orcid.org/0009-0002-4066-7929"},"institutions":[{"id":"https://openalex.org/I183067930","display_name":"Shanghai Jiao Tong University","ror":"https://ror.org/0220qvk04","country_code":"CN","type":"education","lineage":["https://openalex.org/I183067930"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zelin Peng","raw_affiliation_strings":["Shanghai Jiao Tong University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Shanghai Jiao Tong University","institution_ids":["https://openalex.org/I183067930"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5129688481","display_name":"yu huang","orcid":null},"institutions":[{"id":"https://openalex.org/I183067930","display_name":"Shanghai Jiao Tong University","ror":"https://ror.org/0220qvk04","country_code":"CN","type":"education","lineage":["https://openalex.org/I183067930"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yu Huang","raw_affiliation_strings":["Shanghai Jiao Tong University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Shanghai Jiao Tong University","institution_ids":["https://openalex.org/I183067930"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5129718626","display_name":"Wei Shen","orcid":null},"institutions":[{"id":"https://openalex.org/I183067930","display_name":"Shanghai Jiao Tong University","ror":"https://ror.org/0220qvk04","country_code":"CN","type":"education","lineage":["https://openalex.org/I183067930"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Wei Shen","raw_affiliation_strings":["Shanghai Jiao Tong University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Shanghai Jiao Tong University","institution_ids":["https://openalex.org/I183067930"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I183067930"],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":{"value":0.24145299,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":"40","issue":"13","first_page":"10593","last_page":"10602"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11714","display_name":"Multimodal Machine Learning Applications","score":0.3815000057220459,"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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.3815000057220459,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.1031000018119812,"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/T10028","display_name":"Topic Modeling","score":0.07599999755620956,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/security-token","display_name":"Security token","score":0.7548999786376953},{"id":"https://openalex.org/keywords/segmentation","display_name":"Segmentation","score":0.6013000011444092},{"id":"https://openalex.org/keywords/pruning","display_name":"Pruning","score":0.5443000197410583},{"id":"https://openalex.org/keywords/image-segmentation","display_name":"Image segmentation","score":0.486299991607666},{"id":"https://openalex.org/keywords/image","display_name":"Image (mathematics)","score":0.45419999957084656},{"id":"https://openalex.org/keywords/object","display_name":"Object (grammar)","score":0.44269999861717224},{"id":"https://openalex.org/keywords/computation","display_name":"Computation","score":0.4108000099658966},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.37299999594688416},{"id":"https://openalex.org/keywords/abstraction","display_name":"Abstraction","score":0.3513999879360199}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8393999934196472},{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.7548999786376953},{"id":"https://openalex.org/C89600930","wikidata":"https://www.wikidata.org/wiki/Q1423946","display_name":"Segmentation","level":2,"score":0.6013000011444092},{"id":"https://openalex.org/C108010975","wikidata":"https://www.wikidata.org/wiki/Q500094","display_name":"Pruning","level":2,"score":0.5443000197410583},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5231000185012817},{"id":"https://openalex.org/C124504099","wikidata":"https://www.wikidata.org/wiki/Q56933","display_name":"Image segmentation","level":3,"score":0.486299991607666},{"id":"https://openalex.org/C115961682","wikidata":"https://www.wikidata.org/wiki/Q860623","display_name":"Image (mathematics)","level":2,"score":0.45419999957084656},{"id":"https://openalex.org/C2781238097","wikidata":"https://www.wikidata.org/wiki/Q175026","display_name":"Object (grammar)","level":2,"score":0.44269999861717224},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.4108000099658966},{"id":"https://openalex.org/C31972630","wikidata":"https://www.wikidata.org/wiki/Q844240","display_name":"Computer vision","level":1,"score":0.3977999985218048},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.37299999594688416},{"id":"https://openalex.org/C124304363","wikidata":"https://www.wikidata.org/wiki/Q673661","display_name":"Abstraction","level":2,"score":0.3513999879360199},{"id":"https://openalex.org/C74172769","wikidata":"https://www.wikidata.org/wiki/Q1446839","display_name":"Routing (electronic design automation)","level":2,"score":0.33959999680519104},{"id":"https://openalex.org/C160633673","wikidata":"https://www.wikidata.org/wiki/Q355198","display_name":"Pixel","level":2,"score":0.32339999079704285},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.3208000063896179},{"id":"https://openalex.org/C2777735758","wikidata":"https://www.wikidata.org/wiki/Q817765","display_name":"Path (computing)","level":2,"score":0.31369999051094055},{"id":"https://openalex.org/C2776401178","wikidata":"https://www.wikidata.org/wiki/Q12050496","display_name":"Feature (linguistics)","level":2,"score":0.2987000048160553},{"id":"https://openalex.org/C108583219","wikidata":"https://www.wikidata.org/wiki/Q197536","display_name":"Deep learning","level":2,"score":0.27309998869895935},{"id":"https://openalex.org/C184337299","wikidata":"https://www.wikidata.org/wiki/Q1437428","display_name":"Semantics (computer science)","level":2,"score":0.2700999975204468},{"id":"https://openalex.org/C38858127","wikidata":"https://www.wikidata.org/wiki/Q5441228","display_name":"Feed forward","level":2,"score":0.26499998569488525},{"id":"https://openalex.org/C165696696","wikidata":"https://www.wikidata.org/wiki/Q11287","display_name":"Exploit","level":2,"score":0.2630999982357025},{"id":"https://openalex.org/C52622490","wikidata":"https://www.wikidata.org/wiki/Q1026626","display_name":"Feature extraction","level":2,"score":0.2630000114440918},{"id":"https://openalex.org/C9417928","wikidata":"https://www.wikidata.org/wiki/Q1070689","display_name":"Image processing","level":3,"score":0.260699987411499},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.25119999051094055},{"id":"https://openalex.org/C22367795","wikidata":"https://www.wikidata.org/wiki/Q7625208","display_name":"Structured prediction","level":2,"score":0.25110000371932983},{"id":"https://openalex.org/C65885262","wikidata":"https://www.wikidata.org/wiki/Q7429708","display_name":"Scale-space segmentation","level":4,"score":0.250900000333786}],"mesh":[],"locations_count":2,"locations":[{"id":"doi:10.1609/aaai.v40i13.38032","is_oa":true,"landing_page_url":"https://doi.org/10.1609/aaai.v40i13.38032","pdf_url":"https://ojs.aaai.org/index.php/AAAI/article/download/38032/41994","source":{"id":"https://openalex.org/S4210191458","display_name":"Proceedings of the AAAI Conference on Artificial Intelligence","issn_l":"2159-5399","issn":["2159-5399","2374-3468"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/P4310320058","host_organization_name":"Association for the Advancement of Artificial Intelligence","host_organization_lineage":["https://openalex.org/P4310320058"],"host_organization_lineage_names":["Association for the Advancement of Artificial Intelligence"],"type":"conference"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the AAAI Conference on Artificial Intelligence","raw_type":"journal-article"},{"id":"pmh:oai:ojs.aaai.org:article/38032","is_oa":false,"landing_page_url":"https://ojs.aaai.org/index.php/AAAI/article/view/38032","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"2159-5399","raw_type":"info:eu-repo/semantics/article"}],"best_oa_location":{"id":"doi:10.1609/aaai.v40i13.38032","is_oa":true,"landing_page_url":"https://doi.org/10.1609/aaai.v40i13.38032","pdf_url":"https://ojs.aaai.org/index.php/AAAI/article/download/38032/41994","source":{"id":"https://openalex.org/S4210191458","display_name":"Proceedings of the AAAI Conference on Artificial Intelligence","issn_l":"2159-5399","issn":["2159-5399","2374-3468"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/P4310320058","host_organization_name":"Association for the Advancement of Artificial Intelligence","host_organization_lineage":["https://openalex.org/P4310320058"],"host_organization_lineage_names":["Association for the Advancement of Artificial Intelligence"],"type":"conference"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the AAAI Conference on Artificial Intelligence","raw_type":"journal-article"},"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G6402690117","display_name":null,"funder_award_id":"62322604","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"}],"funders":[{"id":"https://openalex.org/F4320321001","display_name":"National Natural Science Foundation of China","ror":"https://ror.org/01h0zpd94"}],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7138011493.pdf","grobid_xml":"https://content.openalex.org/works/W7138011493.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Recent":[0],"advances":[1],"in":[2,12,31,54,104,170,196],"multimodal":[3],"large":[4],"language":[5],"models":[6],"(MLLMs)":[7],"have":[8],"demonstrated":[9],"strong":[10],"capabilities":[11],"addressing":[13],"open-world":[14],"segmentation":[15,32,148,268],"tasks.":[16],"However,":[17],"the":[18,23,55,62,65,87,99,108,142,183,219,222,239,242,254,266],"substantial":[19],"computational":[20],"cost":[21,49],"of":[22,64,89,101,111,182,206,221,241],"LLM":[24,184,256],"components":[25],"presents":[26],"a":[27,39,155,203,234,250],"significant":[28],"challenge,":[29],"especially":[30],"tasks,":[33],"where":[34],"efficiency":[35],"has":[36],"long":[37],"been":[38],"central":[40],"concern.":[41],"Existing":[42],"efficient":[43,147],"MLLM":[44],"approaches":[45],"typically":[46],"reduce":[47,186],"computation":[48],"by":[50,153,227,258],"pruning":[51],"visual":[52,83,102],"tokens":[53,84,103,116,123,131,136,168,195,224],"early":[56,197],"layers,":[57,128],"as":[58,79],"they":[59,175,190],"account":[60],"for":[61],"majority":[63],"input":[66],"sequence.":[67],"Despite":[68],"their":[69,245],"efficiency,":[70],"this":[71],"is":[72,225],"incompatible":[73],"with":[74],"dense":[75],"prediction":[76],"tasks":[77],"such":[78],"segmentation,":[80,105],"since":[81,189],"removing":[82],"leads":[85],"to":[86,134,185,193,262],"loss":[88],"essential":[90],"object":[91],"parts":[92],"and":[93,114,165],"spatial":[94],"details.":[95],"To":[96],"better":[97],"understand":[98],"roles":[100],"we":[106,144],"analyze":[107],"attention":[109],"weights":[110],"both":[112],"image":[113,122,135,194,200],"mask":[115,130,173],"within":[117],"LLM.":[118],"We":[119],"find":[120],"that":[121],"are":[124,176,210],"important":[125],"throughout":[126],"all":[127],"whereas":[129],"only":[132,177,202],"attend":[133,192],"at":[137,179],"deeper":[138,180],"layers.":[139],"Based":[140],"on":[141,151,238],"observation,":[143],"build":[145],"an":[146],"framework":[149],"based":[150],"MLLMs":[152],"introducing":[154],"sophisticated":[156],"token":[157],"routing":[158],"strategy.":[159],"This":[160],"strategy":[161],"dynamically":[162],"determines":[163],"when":[164],"how":[166],"different":[167],"participate":[169],"computation:":[171],"For":[172,199],"tokens,":[174,201],"inserted":[178],"layers":[181],"redundant":[187],"computation,":[188,217],"rarely":[191],"layers;":[198],"small":[204],"number":[205],"them,":[207],"named":[208],"proxies,":[209,229],"updated":[211],"via":[212],"full":[213],"feedforward":[214],"network":[215],"(FFN)":[216],"while":[218,264],"update":[220],"remaining":[223],"guided":[226],"these":[228],"i.e.,":[230],"efficiently":[231],"computed":[232],"through":[233],"lightweight":[235],"projector":[236],"applied":[237],"difference":[240],"proxies":[243],"during":[244],"update.":[246],"Our":[247],"method":[248],"achieves":[249],"1.5\u00d7":[251],"acceleration":[252],"over":[253],"original":[255],"process":[257],"reducing":[259],"its":[260],"FLOPs":[261],"56%,":[263],"maintaining":[265],"same":[267],"performance.":[269]},"counts_by_year":[],"updated_date":"2026-07-02T09:51:11.867554","created_date":"2026-03-18T00:00:00"}