Sign in to view source links and access this dataset
Description
CII-Bench is a multimodal dataset containing 698 Chinese images paired with 800 multiple-choice questions. The images span six domains: Life, Art, Society, Environment, Politics, and Chinese Traditional Culture, and include types such as Illustrations, Memes, Posters, and Comics. It was created by author m-a-p and was last updated on Hugging Face in October 2024.
Use Cases
Benchmarking visual question answering models based on Chinese cultural and societal imagery.
Training multimodal AI on diverse image types like memes, posters, and comics mentioned in the description.
Evaluating model performance across distinct thematic domains such as Politics, Art, and Environment.
Developing culturally-aware AI assistants using images related to Chinese Traditional Culture.
Strengths
Contains 698 distinct images, providing a visual corpus.
Includes 800 multiple-choice questions for structured evaluation.
Covers six distinct thematic domains, suggesting breadth in subject matter.
Features a variety of image types, including illustrations, memes, and comics.
Limitations
Row count and dataset size are unknown, which may limit suitability assessment.
Column-level documentation is absent; field semantics must be inferred after download.
Description metadata is limited; actual data quality requires manual inspection after download.
Provenance
Source
huggingface
Freshness
Last updated 2024-10-18 08:21:35; freshness should be verified.
Geography
The dataset focuses on Chinese images and cultural content.
License is unknown; users should verify terms of use before application.