Divotion's VSRQA Database contains original Full HD uncompressed videos and compressed variants for video quality assessment. The dataset supports research on Super Resolution and Video Compression, with folders organized by sequence name, scale factor, codec name, and bitrate/QP values.
Use Cases
- Train a video quality assessment model using compressed video frames from the GTCompressed folder with scale factor and codec name as features.
- Analyze the impact of target bitrate or qp values on perceptual quality across different video codecs.
- Benchmark Super Resolution algorithms by comparing upscaled outputs against original GT video sequences.
- Study compression artifact patterns using the GTCompressedFullHD folder's reduced-resolution video frames.
Strengths
- Provides structured folders for original (GT) and compressed (GTCompressed) video sequences.
- Includes multiple compression conditions defined by scale factor, codec name, and target bitrate or qp values.
- Dataset is actively maintained, with a last update recorded in September 2025.
Limitations
- Dataset size, row count, and specific file formats are unknown, hindering scalability assessment.
- Lacks detailed column definitions or sample data, requiring users to infer structure from folder names.
- Potential bias from unknown video source content and geographic origin.
Provenance
- Source
- Divotion on Hugging Face
- Collection Method
- Collection of original Full HD videos and their compressed derivatives using various video codecs.
- Time Range
- null
- Freshness
- 2025-09-23
- Geography
- null