EgMM-Corpus: A Multimodal Vision-Language Dataset for Egyptian Culture

2025-10-17

AI Summary

To address the underrepresentation of Middle Eastern and African cultural semantics and the strong Western bias prevalent in existing multimodal datasets, this work introduces EGY-VL—the first high-quality, Egypt-specific vision-language dataset. It comprises over 3,000 culturally authentic images annotated with 313 indigenous concepts spanning landmarks, cuisine, folklore, and traditional practices; all samples undergo dual human verification for cultural fidelity and image–text alignment. Zero-shot classification evaluation on EGY-VL reveals that CLIP achieves only 21.2% Top-1 accuracy, exposing a critical systemic gap in mainstream foundation models’ non-Western cultural understanding. EGY-VL thus establishes the first regional cultural benchmark for multimodal AI evaluation and provides a rigorous, community-grounded standard to assess and advance cultural inclusivity in vision-language models.

Technology Category

• Computer Vision: Language and Vision
• Machine Learning: Multimodal Learning
• Natural Language Processing: Language Grounding & Multi-modal NLP

Application Category

• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• Web Mining and Content Analysis: Mining multimedia, multimodal, multilingual, cross-lingual Web data
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets

Abstract

Despite recent advances in AI, multimodal culturally diverse datasets are still limited, particularly for regions in the Middle East and Africa. In this paper, we introduce EgMM-Corpus, a multimodal dataset dedicated to Egyptian culture. By designing and running a new data collection pipeline, we collected over 3,000 images, covering 313 concepts across landmarks, food, and folklore. Each entry in the dataset is manually validated for cultural authenticity and multimodal coherence. EgMM-Corpus aims to provide a reliable resource for evaluating and training vision-language models in an Egyptian cultural context. We further evaluate the zero-shot performance of Contrastive Language-Image Pre-training CLIP on EgMM-Corpus, on which it achieves 21.2% Top-1 accuracy and 36.4% Top-5 accuracy in classification. These results underscore the existing cultural bias in large-scale vision-language models and demonstrate the importance of EgMM-Corpus as a benchmark for developing culturally aware models.

Problem

• Addressing limited multimodal datasets for Middle Eastern cultures
• Providing culturally authentic Egyptian vision-language dataset
• Evaluating cultural bias in vision-language models

Innovation

• Designed new data collection pipeline for multimodal dataset
• Collected 3000 culturally authentic Egyptian images manually
• Evaluated CLIP model performance on Egyptian cultural benchmark