Arabic Image Captioning 100M: Large-Scale Multimodal Dataset

100 million Arabic image captions form the first large-scale multimodal resource for the Arabic language. Generated using the Mutarjim translation model, this dataset addresses a critical gap in Arabic multimodal AI resources. The dataset was created by Misraj and last updated on May 27, 2025.

Use Cases

• Training Arabic image captioning models based on the 100 million image-text pairs.
• Developing Arabic visual question answering systems based on the multimodal structure.
• Benchmarking Arabic vision-language model performance based on the dataset's scale.
• Fine-tuning translation models for image-related text based on the Mutarjim translation method mentioned.

Strengths

• Contains 100 million image-caption pairs, representing a large-scale resource.
• Described as the first comprehensive Arabic multimodal dataset of this scale and quality.
• Addresses a critical gap in Arabic multimodal AI resources.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Data may reflect translation bias inherent to the generation method.

Provenance

Source

huggingface

Collection Method

Generated using the Mutarjim translation model.

Freshness

Last updated 2025-05-27 12:47:43; freshness should be verified.