Publication:
Crowdsource, Crawl, or Generate? Creating SEA-VL, a Multicultural Vision-Language Dataset for Southeast Asia

3
0

Issued Date

2025-01-01

Resource Type

Conference Paper

ISSN

0736587X

Scopus ID

2-s2.0-105021028710

Journal Title

Proceedings of the Annual Meeting of the Association for Computational Linguistics

Volume

1

Start Page

18685

End Page

18717

Rights Holder(s)

SCOPUS

Bibliographic Citation

Proceedings of the Annual Meeting of the Association for Computational Linguistics Vol.1 (2025) , 18685-18717

Suggested Citation

Cahyawijaya S., Lovenia H., Moniz J.R.A., Wong T.H., Farhansyah M.R., Maung T.T., Hudi F., Anugraha D., Habibi M.R.S., Qorib M.R., Agarwal A., Imperial J.M., Patel H.L., Feliren V., Nasution B.I., Rufino M.A., Winata G.I., Rajagede R.A., Catalan C.R., Imam M.F., Pattnayak P., Pranida S.Z., Pratama K., Bangera Y., Na-Thalang A., Monderin P.N., Song Y., Simon C., Ng L.H.X., Sapan R.L., Rafi T.H., Wang B., Supryadi, Veerakanjana K., Ittichaiwong P., Roque M.T., Vincentio K., Kreangphet T., Artkaew P., Palgunadi K.H., Yu Y., Hastuti R.P., Nixon W., Bangera M., Lim A.X.W., Khine A.H., Zhafran H.M., Ferdinan T., Izzani A.A., Singh A., Evan, Krito J.A., Anugraha M., Ilasariya F.A., Li H., Daniswara J.A., Tjiaranata F.A., Yulianrifat E.P., Udomcharoenchaikit C., Ansori F.R., Ihsani M.K., Nguyen G., Barik A.M., Velasco D.J., Genadi R.A., Saha S., Wei C., Flores I., Chen K.K.H., Santos A.G., Lim W.S., Phyo K.S., Santos T., Dwiastuti M., Luo J., Cruz J.C.B., Hee M.S., Hanif I.A., Alif Al Hakim M., Sya'Ban M.R., Kerdthaisong K., Miranda L.J.V., Koto F., Fatyanosa T.N., Aji A.F., Rosal J.J., Kevin J., Wijaya R., Kampman O.P., Zhang R., Karlsson B.F., Limkonchotiwat P. Crowdsource, Crawl, or Generate? Creating SEA-VL, a Multicultural Vision-Language Dataset for Southeast Asia. Proceedings of the Annual Meeting of the Association for Computational Linguistics Vol.1 (2025) , 18685-18717. 18717. Retrieved from: https://hdl.handle.net/20.500.14740/51577

Title

Crowdsource, Crawl, or Generate? Creating SEA-VL, a Multicultural Vision-Language Dataset for Southeast Asia

Author(s)

Cahyawijaya S.
 Lovenia H.
 Moniz J.R.A.
 Wong T.H.
 Farhansyah M.R.
 Maung T.T.
 Hudi F.
 Anugraha D.
 Habibi M.R.S.
 Qorib M.R.
 Agarwal A.
 Imperial J.M.
 Patel H.L.
 Feliren V.
 Nasution B.I.
 Rufino M.A.
 Winata G.I.
 Rajagede R.A.
 Catalan C.R.
 Imam M.F.
 Pattnayak P.
 Pranida S.Z.
 Pratama K.
 Bangera Y.
 Na-Thalang A.
 Monderin P.N.
 Song Y.
 Simon C.
 Ng L.H.X.
 Sapan R.L.
 Rafi T.H.
 Wang B.
 Supryadi
 Veerakanjana K.
 Ittichaiwong P.
 Roque M.T.
 Vincentio K.
 Kreangphet T.
 Artkaew P.
 Palgunadi K.H.
 Yu Y.
 Hastuti R.P.
 Nixon W.
 Bangera M.
 Lim A.X.W.
 Khine A.H.
 Zhafran H.M.
 Ferdinan T.
 Izzani A.A.
 Singh A.
 Evan
 Krito J.A.
 Anugraha M.
 Ilasariya F.A.
 Li H.
 Daniswara J.A.
 Tjiaranata F.A.
 Yulianrifat E.P.
 Udomcharoenchaikit C.
 Ansori F.R.
 Ihsani M.K.
 Nguyen G.
 Barik A.M.
 Velasco D.J.
 Genadi R.A.
 Saha S.
 Wei C.
 Flores I.
 Chen K.K.H.
 Santos A.G.
 Lim W.S.
 Phyo K.S.
 Santos T.
 Dwiastuti M.
 Luo J.
 Cruz J.C.B.
 Hee M.S.
 Hanif I.A.
 Alif Al Hakim M.
 Sya'Ban M.R.
 Kerdthaisong K.
 Miranda L.J.V.
 Koto F.
 Fatyanosa T.N.
 Aji A.F.
 Rosal J.J.
 Kevin J.
 Wijaya R.
 Kampman O.P.
 Zhang R.
 Karlsson B.F.
 Limkonchotiwat P.

Author's Affiliation

University of Toronto
 University of Illinois Urbana-Champaign
 The University of Manchester
 National University of Singapore
 Monash University
 Tianjin University
 New York University
 Carnegie Mellon University
 Brown University
 Auburn University
 Hanyang University
 Chulalongkorn University
 University of Bath
 Universitas Indonesia
 Polytechnique Montréal
 Universitas Gadjah Mada
 Institut Teknologi Bandung
 Nara Institute of Science and Technology
 Thammasat University
 Institut Teknologi Sepuluh Nopember
 Macau University of Science and Technology
 Brawijaya University
 Siriraj Hospital
 King Mongkut's University of Technology Thonburi
 Bina Nusantara University
 Indian Statistical Institute, Kolkata
 Ton-Duc-Thang University
 Seoul National University of Science and Technology
 A-Star, Institute for Infocomm Research
 Singapore University of Technology and Design
 Srinakharinwirot University
 Universitas Islam Indonesia
 Ateneo de Manila University
 University of New Haven
 Mohamed Bin Zayed University of Artificial Intelligence
 Montreal Institute for Learning Algorithms
 Universitas Pelita Harapan
 Oracle Corporation
 University of the Philippines
 Vidyasirimedhi Institute of Science and Technology
 Singapore Polytechnic
 National University, Philippines
 Sony Group Corporation
 Graphcore Limited
 MOH Office for Healthcare Transformation
 Allen Institute for AI
 AI Singapore
 Beijing Academy of Artificial Intelligence (BAAI)
 Wroclaw Tech
 Cohere
 SCB 10X
 Meta
 Samsung R&D Institute Philippines
 Capital One
 Works Applications Lab
 SEACrowd
 Dataxet:Sonar
 IndoNLP

Corresponding Author(s)

Cahyawijaya S.

Other Contributor(s)

Srinakharinwirot University

Abstract

Despite Southeast Asia's (SEA) extraordinary linguistic and cultural diversity, the region remains significantly underrepresented in vision-language (VL) research, resulting in AI models that inadequately capture SEA cultural nuances. To fill this gap, we present SEA-VL, an open-source initiative dedicated to developing culturally relevant high-quality datasets for SEA languages. By involving contributors from SEA countries, SEA-VL ensures better cultural relevance and diversity, fostering greater inclusivity of underrepresented languages and cultural depictions in VL research. Our methodology employed three approaches: community-driven crowdsourcing with SEA contributors, automated image crawling, and synthetic image generation. We evaluated each method's effectiveness in capturing cultural relevance. We found that image crawling achieves approximately ∼85% cultural relevance while being more cost- and time-efficient than crowdsourcing, whereas synthetic image generation failed to accurately reflect SEA cultural nuances and contexts. Collectively, we gathered 1.28 million SEA culturally relevant images, more than 50 times larger than other existing datasets. This work bridges the representation gap in SEA, establishes a foundation for developing culturally aware AI systems for this region, and provides a replicable framework for addressing representation gaps in other underrepresented regions.

Keyword(s)

Social Sciences
 Computer Science
 Arts and Humanities

View online Resources

Research Projects

Organizational Units

Journal Issue

URI

https://hdl.handle.net/20.500.14740/51577

Collections

Scopus: Year 2025

Endorsement

Review

Supplemented By

Referenced By