Simala1/echicgv1

Dataset Card for EChICG v1

Dataset Summary

EChICG v1 (Ethiopian Cultural & Historical Image Captioning Dataset) is designed for training and evaluating vision-language models, image captioning architectures, and multimodal AI systems in Amharic.

• Original Source / Kaggle Repository: https://www.kaggle.com/datasets/simachewalamneh/echicgv1
• Language: Amharic (am)
• Total Images Scanned: 4,294
• Unique Images: 4,128 (166 duplicate image variants identified)

Dataset Structure

• Images/: Directory containing cultural and historical image files.
• Amharic CHI caption.txt: Text file containing image-to-caption mappings in Amharic.

Intended Use

• Amharic Image Captioning
• Multimodal Vision-Language Model Fine-Tuning
• Cultural & Historical Visual Understanding Tasks