WORLDCUISINES: A Massive-Scale Benchmark for Multilingual and Multicultural Visual Question Answering on Global Cuisines

Genta Winata; Frederikus Hudi; Patrick Amadeus Irawan; David Anugraha; Rifki Afina Putri; Yutong Wang; Adam Nohejl; Ubaidillah Prathama; Nedjma  Ousidhoum; Afifa Amriani; Anar Rzayev; Anirban Das; Ashmari Pramodya; Aulia Adila; Bryan Willie; Candy Olivia Mawalim; Ching Lam Cheng; Daud Abolade; Emmanuele Chersoni; Enrico Santus; Fariz Ikhwantri; Garry Kuwanto; Hanyang Zhao; Haryo Akbarianto Wibowo; Holy Lovenia; Jan Christian Blaise Cruz; Jan Vira Gotama Putra; Junho Myung; Lucky Susanto; Maria Angelica Riera Machin; Marina Zhukova; Michael Anugraha; Muhammad Farid Adilazuarda; Natasha Santosa; Peerat Limkonchotiwat; Raj Dabre; Rio Alexander Audino; Samuel Cahyawijaya; Shi-Xiong Zhang; Stephanie Yulia Salim; Yi Zhou; Yinxuan Gui; David Ifeoluwa Adelani; En-Shiun Annie Lee; Shogo Okada; Ayu Purwarianti; Alham Fikri Aji; Taro Watanabe; Derry Tanti Wijaya; Alice Oh; Chong-Wah Ngo

WORLDCUISINES: A Massive-Scale Benchmark for Multilingual and Multicultural Visual Question Answering on Global Cuisines

Genta Winata, Frederikus Hudi, Patrick Amadeus Irawan, David Anugraha, Rifki Afina Putri, Yutong Wang, Adam Nohejl, Ubaidillah Prathama, Nedjma Ousidhoum, Afifa Amriani, Anar Rzayev, Anirban Das, Ashmari Pramodya, Aulia Adila, Bryan Willie, Candy Olivia Mawalim, Ching Lam Cheng, Daud Abolade, Emmanuele Chersoni, Enrico SantusFariz Ikhwantri, Garry Kuwanto, Hanyang Zhao, Haryo Akbarianto Wibowo, Holy Lovenia, Jan Christian Blaise Cruz, Jan Vira Gotama Putra, Junho Myung, Lucky Susanto, Maria Angelica Riera Machin, Marina Zhukova, Michael Anugraha, Muhammad Farid Adilazuarda, Natasha Santosa, Peerat Limkonchotiwat, Raj Dabre, Rio Alexander Audino, Samuel Cahyawijaya, Shi-Xiong Zhang, Stephanie Yulia Salim, Yi Zhou, Yinxuan Gui, David Ifeoluwa Adelani, En-Shiun Annie Lee, Shogo Okada, Ayu Purwarianti, Alham Fikri Aji, Taro Watanabe, Derry Tanti Wijaya, Alice Oh, Chong-Wah Ngo

Department of Chinese and Bilingual Studies

Research output: Chapter in book / Conference proceeding › Conference article published in proceeding or book › Academic research › peer-review

Abstract

Vision Language Models (VLMs) often struggle with culture-specific knowledge, particularly in languages other than English and in underrepresented cultural contexts. To evaluate their understanding of such knowledge, we introduce WORLDCUISINES, a massivescale benchmark for multilingual and multicultural,
visually grounded language understanding. This benchmark includes a visual question answering (VQA) dataset with text-image pairs across 30 languages and dialects, spanning 9 language families and featuring over 1 million data points, making it the largest multicultural VQA benchmark to date. It includes tasks for identifying dish names and their origins. We provide evaluation datasets in two sizes (12k and 60k instances) alongside a training dataset (1 million instances). Our findings show that while VLMs perform better with correct location context, they struggle with adversarial contexts and predicting specific regional cuisines and languages. To support future research, we release a knowledge base with annotated food entries and images along with the VQA data.

Original language	English
Title of host publication	Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers)
Editors	Luis Chiruzzo, Alan Ritter, Lu Wang
Publisher	Association for Computational Linguistics
Pages	3242-3264
ISBN (Electronic)	9798891761896
Publication status	Published - Apr 2025
Event	2025 Annual Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics - Albuquerque Convention Center, Albuquerque, United States Duration: 29 Apr 2025 → 4 May 2025 https://2025.naacl.org/

Conference

Conference	2025 Annual Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics
Abbreviated title	NAACL 2025
Country/Territory	United States
City	Albuquerque
Period	29/04/25 → 4/05/25
Internet address	https://2025.naacl.org/

More information

Cite this

Winata, G., Hudi, F., Irawan, P. A., Anugraha, D., Putri, R. A., Wang, Y., Nohejl, A., Prathama, U., Ousidhoum, N., Amriani, A., Rzayev, A., Das, A., Pramodya, A., Adila, A., Willie, B., Mawalim, C. O., Cheng, C. L., Abolade, D., Chersoni, E., ... Ngo, C.-W. (2025). WORLDCUISINES: A Massive-Scale Benchmark for Multilingual and Multicultural Visual Question Answering on Global Cuisines. In L. Chiruzzo, A. Ritter, & L. Wang (Eds.), Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers) (pp. 3242-3264). Association for Computational Linguistics. https://arxiv.org/pdf/2410.12705

Winata, Genta ; Hudi, Frederikus ; Irawan, Patrick Amadeus et al. / WORLDCUISINES : A Massive-Scale Benchmark for Multilingual and Multicultural Visual Question Answering on Global Cuisines. Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). editor / Luis Chiruzzo ; Alan Ritter ; Lu Wang. Association for Computational Linguistics, 2025. pp. 3242-3264

@inproceedings{5d68275397514eb980e883d94ca3757b,

title = "WORLDCUISINES: A Massive-Scale Benchmark for Multilingual and Multicultural Visual Question Answering on Global Cuisines",

abstract = "Vision Language Models (VLMs) often struggle with culture-specific knowledge, particularly in languages other than English and in underrepresented cultural contexts. To evaluate their understanding of such knowledge, we introduce WORLDCUISINES, a massivescale benchmark for multilingual and multicultural,visually grounded language understanding. This benchmark includes a visual question answering (VQA) dataset with text-image pairs across 30 languages and dialects, spanning 9 language families and featuring over 1 million data points, making it the largest multicultural VQA benchmark to date. It includes tasks for identifying dish names and their origins. We provide evaluation datasets in two sizes (12k and 60k instances) alongside a training dataset (1 million instances). Our findings show that while VLMs perform better with correct location context, they struggle with adversarial contexts and predicting specific regional cuisines and languages. To support future research, we release a knowledge base with annotated food entries and images along with the VQA data.",

author = "Genta Winata and Frederikus Hudi and Irawan, \{Patrick Amadeus\} and David Anugraha and Putri, \{Rifki Afina\} and Yutong Wang and Adam Nohejl and Ubaidillah Prathama and Nedjma Ousidhoum and Afifa Amriani and Anar Rzayev and Anirban Das and Ashmari Pramodya and Aulia Adila and Bryan Willie and Mawalim, \{Candy Olivia\} and Cheng, \{Ching Lam\} and Daud Abolade and Emmanuele Chersoni and Enrico Santus and Fariz Ikhwantri and Garry Kuwanto and Hanyang Zhao and Wibowo, \{Haryo Akbarianto\} and Holy Lovenia and Cruz, \{Jan Christian Blaise\} and Putra, \{Jan Vira Gotama\} and Junho Myung and Lucky Susanto and Machin, \{Maria Angelica Riera\} and Marina Zhukova and Michael Anugraha and Adilazuarda, \{Muhammad Farid\} and Natasha Santosa and Peerat Limkonchotiwat and Raj Dabre and Audino, \{Rio Alexander\} and Samuel Cahyawijaya and Shi-Xiong Zhang and Salim, \{Stephanie Yulia\} and Yi Zhou and Yinxuan Gui and Adelani, \{David Ifeoluwa\} and Lee, \{En-Shiun Annie\} and Shogo Okada and Ayu Purwarianti and Aji, \{Alham Fikri\} and Taro Watanabe and Wijaya, \{Derry Tanti\} and Alice Oh and Chong-Wah Ngo",

year = "2025",

month = apr,

language = "English",

pages = "3242--3264",

editor = "Luis Chiruzzo and Alan Ritter and Lu Wang",

booktitle = "Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers)",

publisher = "Association for Computational Linguistics",

address = "United States",

note = "2025 Annual Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics, NAACL 2025 ; Conference date: 29-04-2025 Through 04-05-2025",

url = "https://2025.naacl.org/",

}

Winata, G, Hudi, F, Irawan, PA, Anugraha, D, Putri, RA, Wang, Y, Nohejl, A, Prathama, U, Ousidhoum, N, Amriani, A, Rzayev, A, Das, A, Pramodya, A, Adila, A, Willie, B, Mawalim, CO, Cheng, CL, Abolade, D, Chersoni, E, Santus, E, Ikhwantri, F, Kuwanto, G, Zhao, H, Wibowo, HA, Lovenia, H, Cruz, JCB, Putra, JVG, Myung, J, Susanto, L, Machin, MAR, Zhukova, M, Anugraha, M, Adilazuarda, MF, Santosa, N, Limkonchotiwat, P, Dabre, R, Audino, RA, Cahyawijaya, S, Zhang, S-X, Salim, SY, Zhou, Y, Gui, Y, Adelani, DI, Lee, E-SA, Okada, S, Purwarianti, A, Aji, AF, Watanabe, T, Wijaya, DT, Oh, A & Ngo, C-W 2025, WORLDCUISINES: A Massive-Scale Benchmark for Multilingual and Multicultural Visual Question Answering on Global Cuisines. in L Chiruzzo, A Ritter & L Wang (eds), Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). Association for Computational Linguistics, pp. 3242-3264, 2025 Annual Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics, Albuquerque, New Mexico, United States, 29/04/25. <https://arxiv.org/pdf/2410.12705>

WORLDCUISINES: A Massive-Scale Benchmark for Multilingual and Multicultural Visual Question Answering on Global Cuisines. / Winata, Genta; Hudi, Frederikus; Irawan, Patrick Amadeus et al.
Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). ed. / Luis Chiruzzo; Alan Ritter; Lu Wang. Association for Computational Linguistics, 2025. p. 3242-3264.

Research output: Chapter in book / Conference proceeding › Conference article published in proceeding or book › Academic research › peer-review

TY - GEN

T1 - WORLDCUISINES

T2 - 2025 Annual Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics

AU - Winata, Genta

AU - Hudi, Frederikus

AU - Irawan, Patrick Amadeus

AU - Anugraha, David

AU - Putri, Rifki Afina

AU - Wang, Yutong

AU - Nohejl, Adam

AU - Prathama, Ubaidillah

AU - Ousidhoum, Nedjma

AU - Amriani, Afifa

AU - Rzayev, Anar

AU - Das, Anirban

AU - Pramodya, Ashmari

AU - Adila, Aulia

AU - Willie, Bryan

AU - Mawalim, Candy Olivia

AU - Cheng, Ching Lam

AU - Abolade, Daud

AU - Chersoni, Emmanuele

AU - Santus, Enrico

AU - Ikhwantri, Fariz

AU - Kuwanto, Garry

AU - Zhao, Hanyang

AU - Wibowo, Haryo Akbarianto

AU - Lovenia, Holy

AU - Cruz, Jan Christian Blaise

AU - Putra, Jan Vira Gotama

AU - Myung, Junho

AU - Susanto, Lucky

AU - Machin, Maria Angelica Riera

AU - Zhukova, Marina

AU - Anugraha, Michael

AU - Adilazuarda, Muhammad Farid

AU - Santosa, Natasha

AU - Limkonchotiwat, Peerat

AU - Dabre, Raj

AU - Audino, Rio Alexander

AU - Cahyawijaya, Samuel

AU - Zhang, Shi-Xiong

AU - Salim, Stephanie Yulia

AU - Zhou, Yi

AU - Gui, Yinxuan

AU - Adelani, David Ifeoluwa

AU - Lee, En-Shiun Annie

AU - Okada, Shogo

AU - Purwarianti, Ayu

AU - Aji, Alham Fikri

AU - Watanabe, Taro

AU - Wijaya, Derry Tanti

AU - Oh, Alice

AU - Ngo, Chong-Wah

PY - 2025/4

Y1 - 2025/4

N2 - Vision Language Models (VLMs) often struggle with culture-specific knowledge, particularly in languages other than English and in underrepresented cultural contexts. To evaluate their understanding of such knowledge, we introduce WORLDCUISINES, a massivescale benchmark for multilingual and multicultural,visually grounded language understanding. This benchmark includes a visual question answering (VQA) dataset with text-image pairs across 30 languages and dialects, spanning 9 language families and featuring over 1 million data points, making it the largest multicultural VQA benchmark to date. It includes tasks for identifying dish names and their origins. We provide evaluation datasets in two sizes (12k and 60k instances) alongside a training dataset (1 million instances). Our findings show that while VLMs perform better with correct location context, they struggle with adversarial contexts and predicting specific regional cuisines and languages. To support future research, we release a knowledge base with annotated food entries and images along with the VQA data.

AB - Vision Language Models (VLMs) often struggle with culture-specific knowledge, particularly in languages other than English and in underrepresented cultural contexts. To evaluate their understanding of such knowledge, we introduce WORLDCUISINES, a massivescale benchmark for multilingual and multicultural,visually grounded language understanding. This benchmark includes a visual question answering (VQA) dataset with text-image pairs across 30 languages and dialects, spanning 9 language families and featuring over 1 million data points, making it the largest multicultural VQA benchmark to date. It includes tasks for identifying dish names and their origins. We provide evaluation datasets in two sizes (12k and 60k instances) alongside a training dataset (1 million instances). Our findings show that while VLMs perform better with correct location context, they struggle with adversarial contexts and predicting specific regional cuisines and languages. To support future research, we release a knowledge base with annotated food entries and images along with the VQA data.

M3 - Conference article published in proceeding or book

SP - 3242

EP - 3264

BT - Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers)

A2 - Chiruzzo, Luis

A2 - Ritter, Alan

A2 - Wang, Lu

PB - Association for Computational Linguistics

Y2 - 29 April 2025 through 4 May 2025

ER -

Winata G, Hudi F, Irawan PA, Anugraha D, Putri RA, Wang Y et al. WORLDCUISINES: A Massive-Scale Benchmark for Multilingual and Multicultural Visual Question Answering on Global Cuisines. In Chiruzzo L, Ritter A, Wang L, editors, Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). Association for Computational Linguistics. 2025. p. 3242-3264

WORLDCUISINES: A Massive-Scale Benchmark for Multilingual and Multicultural Visual Question Answering on Global Cuisines

Abstract

Conference

More information

Fingerprint

Cite this