SciOL-CI: A Large-Scale Multimodal Scientific Dataset for Image-Text Tasks

Scientific Openly-Licensed Publications (SciOL) and its companion dataset, MuLMS-Img, are introduced in a WACV 2024 paper by Tim Tarsi et al. The dataset is designed for image-text tasks within the scientific domain and is hosted on HuggingFace by the author Timbrt. The dataset page was last updated on April 17, 2024.

Use Cases

• Training multimodal vision-language models based on scientific image-text pairs.
• Benchmarking model performance on scientific image-text tasks as described in the WACV 2024 paper.
• Conducting research on scientific document understanding using openly-licensed publications.

Strengths

• Dataset is associated with a peer-reviewed publication (WACV 2024).
• Designed specifically for image-text tasks in the scientific domain.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count and dataset size are unknown, which may limit suitability assessment.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

Timbrt (likely Tim Tarsi) on HuggingFace.

Collection Method

Companion material for a scientific publication; likely compiled from openly-licensed scientific publications.

Freshness

Last updated 2024-04-17 18:47:42; freshness should be verified.