Formosa Vision: Taiwan-Centric Visual Language Dataset with Community-Curated Descriptions

Formosa Vision is an open-source visual language dataset focused on Taiwanese local culture, containing over two thousand images selected from the National Cultural Memory Bank 2.0. The dataset was created by the Twinkle AI community using a hybrid method where visual language models generated image dialogues, which were then manually checked and revised by participants. It was last updated on November 20, 2025.

Use Cases

• Training vision-language models on Taiwanese cultural imagery based on the described scenes and historical contexts.
• Fine-tuning models for generating culturally consistent image captions in Traditional Chinese.
• Developing multimodal AI applications that require understanding of Taiwanese landmarks, architecture, and daily life.

Strengths

• Contains over two thousand images with associated text descriptions.
• Descriptions were generated by a model and then manually reviewed and revised for accuracy and naturalness.
• Images are sourced from the National Cultural Memory Bank 2.0 and are licensed under CC BY-SA or more open terms.

Limitations

• Row count is unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• Data may reflect geographic and cultural bias inherent to its specific source.

Provenance

Source

Twinkle AI community, sourcing images from the National Cultural Memory Bank 2.0.

Collection Method

Model-generated image dialogues followed by manual participant review and revision.

Freshness

Last updated 2025-11-20 01:20:50.

Geography

Taiwan