RealQA Multimodal Dataset for Image Quality Assessment

The RealQA dataset is proposed for User Generated Content (UGC) assessment, while other real-world captured datasets like AVA, TAD66K, Koniq-10k, livec, and spaq are used for IQA and IAA. It contains multimodal data, including images and text, as indicated by its tags. The dataset was created by MingxingLi and was last updated on May 29, 2025.

Use Cases

• Train a multimodal model for Image Quality Assessment using the provided image and text data.
• Analyze User Generated Content quality by leveraging the dataset's focus on UGC assessment.
• Benchmark IQA and IAA methods against established datasets like AVA, TAD66K, and Koniq-10k referenced in the description.

Strengths

• Focuses on real-world captured images, as referenced by the use of AVA, TAD66K, Koniq-10k, livec, and spaq datasets.
• Designed for the specific task of User Generated Content (UGC) assessment.
• Multimodal nature includes both Image and Text data, as indicated by the tags.

Limitations

• Specific dataset size, row count, column structure, and file formats are unknown.
• The relationship and exact contents of the RealQA dataset versus the other mentioned datasets (AVA, TAD66K, etc.) are not clearly defined.
• License information is not provided, which is critical for usage.

Provenance

Source

huggingface

Collection Method

Proposed by the author for UGC assessment; other referenced datasets are captured in the real world.

Freshness

Last updated on May 29, 2025.