Doc VQA: Multimodal Question-Answering on Infographics and Figures

The dataset integrates images from the Infographic_vqa and AFTDB (Arxiv Figure Table Database) collections. It consists of image-text pairs, with each image linked to an average of five questions and answers available in both English and French. The dataset was created by cmarkea and last updated on Hugging Face in August 2024.

Use Cases

• Train visual question-answering models based on infographic images.
• Benchmark document understanding systems based on figures and tables from academic papers.
• Develop multilingual (English/French) QA systems based on the parallel question-answer pairs.
• Fine-tune large language models for multimodal reasoning based on the generated Gemini 1.5 Pro QA pairs.

Strengths

• Pairs each image with an average of five questions and answers.
• Questions and answers are available in both English and French.
• Integrates images from two established sources: Infographic_vqa and AFTDB.

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 source bias inherent to the original Infographic_vqa and AFTDB collections.

Provenance

Source

Images sourced from HuggingFaceM4's Infographic_vqa dataset and cmarkea's AFTDB (Arxiv Figure Table Database).

Collection Method

Questions and answers were generated using Gemini 1.5 Pro.

Freshness

Last updated 2024-08-26 14:16:40; freshness should be verified.