{"id":"https://openalex.org/W4416065032","doi":"https://doi.org/10.48550/arxiv.2509.07120","title":"Block-Sparse Global Attention for Efficient Multi-View Geometry Transformers","display_name":"Block-Sparse Global Attention for Efficient Multi-View Geometry Transformers","publication_year":2025,"publication_date":"2025-09-08","ids":{"openalex":"https://openalex.org/W4416065032","doi":"https://doi.org/10.48550/arxiv.2509.07120"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2509.07120","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2509.07120","pdf_url":"https://arxiv.org/pdf/2509.07120","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,"listed_in":[],"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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"volume":null,"type":"preprint","is_traditional":true,"indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2509.07120","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":null,"display_name":"Wang, Chung-Shien Brian","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Chung-Shien Brian","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Schmidt, Christian","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I887968799","display_name":"RWTH Aachen University","ror":"https://ror.org/04xfq0f34","country_code":"DE","type":"education","lineage":["https://openalex.org/I887968799"]}],"countries":["DE"],"is_corresponding":false,"raw_author_name":"Schmidt, Christian","raw_affiliation_strings":["Computer Vision Group RWTH Aachen University https :// vision "],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Computer Vision Group RWTH Aachen University https :// vision ","institution_ids":["https://openalex.org/I887968799"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5107516043","display_name":"Jens Piekenbrinck","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I887968799","display_name":"RWTH Aachen University","ror":"https://ror.org/04xfq0f34","country_code":"DE","type":"education","lineage":["https://openalex.org/I887968799"]}],"countries":["DE"],"is_corresponding":false,"raw_author_name":"Piekenbrinck, Jens","raw_affiliation_strings":["Computer Vision Group RWTH Aachen University https :// vision "],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Computer Vision Group RWTH Aachen University https :// vision ","institution_ids":["https://openalex.org/I887968799"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5071006649","display_name":"Bastian Leibe","orcid":"https://orcid.org/0000-0003-4225-0051","observed_orcids":["https://orcid.org/0000-0003-4225-0051"]},"institutions":[{"id":"https://openalex.org/I887968799","display_name":"RWTH Aachen University","ror":"https://ror.org/04xfq0f34","country_code":"DE","type":"education","lineage":["https://openalex.org/I887968799"]}],"countries":["DE"],"is_corresponding":false,"raw_author_name":"Leibe, Bastian","raw_affiliation_strings":["Computer Vision Group RWTH Aachen University https :// vision "],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Computer Vision Group RWTH Aachen University https :// vision ","institution_ids":["https://openalex.org/I887968799"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"cited_by_count":29,"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/T10531","display_name":"Advanced Vision and Imaging","score":0.7445999979972839,"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/T10531","display_name":"Advanced Vision and Imaging","score":0.7445999979972839,"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/T10036","display_name":"Advanced Neural Network Applications","score":0.10170000046491623,"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/T10627","display_name":"Advanced Image and Video Retrieval Techniques","score":0.09969999641180038,"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/international-attention","display_name":"global attention","score":0.9639999866485596},{"id":"https://openalex.org/keywords/multiple-view-geometry","display_name":"multi-view geometry","score":0.7960000038146973},{"id":"https://openalex.org/keywords/multi-view-reconstruction","display_name":"multi-view reconstruction","score":0.8190000057220459},{"id":"https://openalex.org/keywords/inference-efficiency","display_name":"inference efficiency","score":0.6370000243186951},{"id":"https://openalex.org/keywords/vggt","display_name":"VGGT","score":0.7279999852180481},{"id":"https://openalex.org/keywords/large-image-collections","display_name":"large image collections","score":0.574999988079071}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7511000037193298},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.5810999870300293},{"id":"https://openalex.org/C48044578","wikidata":"https://www.wikidata.org/wiki/Q727490","display_name":"Scalability","level":2,"score":0.5807999968528748},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5047000050544739},{"id":"https://openalex.org/C2779304628","wikidata":"https://www.wikidata.org/wiki/Q3503480","display_name":"Face (sociological concept)","level":2,"score":0.4848000109195709},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.45910000801086426},{"id":"https://openalex.org/C79581498","wikidata":"https://www.wikidata.org/wiki/Q1367530","display_name":"Suite","level":2,"score":0.4578000009059906},{"id":"https://openalex.org/C2780586882","wikidata":"https://www.wikidata.org/wiki/Q7520643","display_name":"Simple (philosophy)","level":2,"score":0.45170000195503235},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.4375},{"id":"https://openalex.org/C129844170","wikidata":"https://www.wikidata.org/wiki/Q41299","display_name":"Quadratic equation","level":2,"score":0.4325000047683716},{"id":"https://openalex.org/C115961682","wikidata":"https://www.wikidata.org/wiki/Q860623","display_name":"Image (mathematics)","level":2,"score":0.3824000060558319},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.3677999973297119},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.3531999886035919},{"id":"https://openalex.org/C179799912","wikidata":"https://www.wikidata.org/wiki/Q205084","display_name":"Computational complexity theory","level":2,"score":0.295199990272522},{"id":"https://openalex.org/C61423126","wikidata":"https://www.wikidata.org/wiki/Q187432","display_name":"Scripting language","level":2,"score":0.29010000824928284},{"id":"https://openalex.org/C175154964","wikidata":"https://www.wikidata.org/wiki/Q380077","display_name":"Task analysis","level":3,"score":0.27090001106262207},{"id":"https://openalex.org/C9417928","wikidata":"https://www.wikidata.org/wiki/Q1070689","display_name":"Image processing","level":3,"score":0.267300009727478},{"id":"https://openalex.org/C2777655017","wikidata":"https://www.wikidata.org/wiki/Q1501161","display_name":"Toolbox","level":2,"score":0.26409998536109924},{"id":"https://openalex.org/C2777472644","wikidata":"https://www.wikidata.org/wiki/Q16968992","display_name":"Approximate inference","level":3,"score":0.2615000009536743},{"id":"https://openalex.org/C31510193","wikidata":"https://www.wikidata.org/wiki/Q1192553","display_name":"Facial recognition system","level":3,"score":0.257099986076355}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2509.07120","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2509.07120","pdf_url":"https://arxiv.org/pdf/2509.07120","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,"listed_in":[],"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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2509.07120","doi":"https://doi.org/10.48550/arxiv.2509.07120","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2509.07120","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,"listed_in":[],"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"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2509.07120","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2509.07120","pdf_url":"https://arxiv.org/pdf/2509.07120","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,"listed_in":[],"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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[{"id":"https://openalex.org/F4320324232","display_name":"RWTH Aachen University","ror":"https://ror.org/04xfq0f34"}],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4416065032.pdf","grobid_xml":"https://content.openalex.org/works/W4416065032.grobid-xml"},"referenced_works_count":12,"referenced_works":["https://openalex.org/W2940744433","https://openalex.org/W3015468748","https://openalex.org/W4366208220","https://openalex.org/W4381586827","https://openalex.org/W4384648639","https://openalex.org/W4403579453","https://openalex.org/W4407764033","https://openalex.org/W4413559025","https://openalex.org/W4414883407","https://openalex.org/W4416063870","https://openalex.org/W4417305659","https://openalex.org/W7082144516"],"related_works":["https://openalex.org/W4417006785","https://openalex.org/W7167218985","https://openalex.org/W4416966127","https://openalex.org/W7165195271","https://openalex.org/W7214778733","https://openalex.org/W7161255291","https://openalex.org/W7221011889","https://openalex.org/W7214042025","https://openalex.org/W7160941174","https://openalex.org/W4387725218"],"abstract_inverted_index":{"Efficient":[0],"and":[1,21,68,91,152],"accurate":[2],"feed-forward":[3],"multi-view":[4,134],"reconstruction":[5],"has":[6],"long":[7],"been":[8],"an":[9],"important":[10],"task":[11,126],"in":[12,96],"computer":[13],"vision.":[14],"Recent":[15],"transformer-based":[16],"models":[17,67],"like":[18],"VGGT,":[19,149],"$\u03c0^3$":[20,150],"MapAnything":[22],"have":[23],"demonstrated":[24],"remarkable":[25],"performance":[26],"with":[27,111],"relatively":[28],"simple":[29],"architectures.":[30],"However,":[31],"their":[32],"scalability":[33,157],"is":[34],"fundamentally":[35],"constrained":[36],"by":[37,93,119],"the":[38,61,71],"quadratic":[39],"complexity":[40],"of":[41,65,79,133],"global":[42,62,108,144],"attention,":[43,109],"which":[44],"imposes":[45],"a":[46,76,102,130],"significant":[47],"runtime":[48],"bottleneck":[49],"when":[50],"processing":[51],"large":[52,97,159],"image":[53,160],"sets.":[54],"In":[55],"this":[56,89],"work,":[57],"we":[58,100],"empirically":[59],"analyze":[60],"attention":[63],"matrix":[64],"these":[66],"observe":[69],"that":[70,137],"probability":[72],"mass":[73],"concentrates":[74],"on":[75,88,129],"small":[77],"subset":[78],"patch-patch":[80],"interactions":[81],"corresponding":[82],"to":[83,158],"cross-view":[84],"geometric":[85],"correspondences.":[86],"Building":[87],"insight":[90],"inspired":[92],"recent":[94],"advances":[95],"language":[98],"models,":[99],"propose":[101],"training-free,":[103],"block-sparse":[104],"replacement":[105],"for":[106],"dense":[107],"implemented":[110],"highly":[112],"optimized":[113],"kernels.":[114],"Our":[115],"method":[116],"accelerates":[117],"inference":[118],"more":[120],"than":[121],"$3\\times$":[122],"while":[123,154],"maintaining":[124],"comparable":[125],"performance.":[127],"Evaluations":[128],"comprehensive":[131],"suite":[132],"benchmarks":[135],"demonstrate":[136],"our":[138],"approach":[139],"seamlessly":[140],"integrates":[141],"into":[142],"existing":[143],"attention-based":[145],"architectures":[146],"such":[147],"as":[148],",":[151],"MapAnything,":[153],"substantially":[155],"improving":[156],"collections.":[161]},"counts_by_year":[{"year":2026,"cited_by_count":24},{"year":2025,"cited_by_count":5}],"updated_date":"2026-11-07T11:25:07.637180","created_date":"2025-10-10T00:00:00"}