Research map / Video / Video-language understanding
Unified Quality Assessment of in-the-Wild Videos with Mixed Datasets Training
Dingquan Li, Tingting Jiang, Ming Feng Jiang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add dataset-specific perceptual scale alignment instead of naive linear re-scaling (ablation) | MDTVSFA mixed training · Four datasets mixed | size-weighted median SROCC: 0.7576 -> 0.7753 | Improved |
| use dataset-specific scale alignment over linear re-scaling (LS-VSFA vs MDTVSFA) (ablation) | MDTVSFA · CVD2014+KoNViD-1k+LIVE-Qualcomm overall | SROCC: 0.7603 -> 0.7860 | Improved |
| use dataset-specific scale alignment over linear re-scaling (LS-VSFA vs MDTVSFA) (ablation) | MDTVSFA · CVD2014+KoNViD-1k+LIVE-Qualcomm overall | PLCC: 0.7662 -> 0.7923 | Improved |
| combine monotonicity, linearity and error losses rather than fewer losses (ablation) | MDTVSFA · Four datasets mixed | size-weighted median SROCC | Improved |
| mix three datasets for training instead of training on KoNViD-1k only (ablation) | Our model · CVD2014+KoNViD-1k+LIVE-Qualcomm overall | median SROCC: 0.7483 -> 0.7829 | Improved |
| mix three datasets for training instead of training on CVD2014 only (ablation) | Our model · CVD2014+KoNViD-1k+LIVE-Qualcomm overall | median SROCC: 0.6165 -> 0.7829 | Improved |
| mix three datasets instead of LIVE-Qualcomm only (ablation) | Our model · LIVE-Qualcomm test | median SROCC: 0.7538 -> 0.8157 | Improved |
| add CVD2014 to KoNViD-1k training, test on KoNViD-1k (ablation) | Our model · KoNViD-1k test | median SROCC: drops 0.0067 | Hurt |
| add CVD2014 to training of KoNViD-1k, test on CVD2014 (ablation) | Our model · CVD2014 test | median SROCC: 0.6474 -> +0.2078 | Improved |
| add LIVE-Qualcomm to LIVE-VQC training, test on unseen CVD2014 and KoNViD-1k (ablation) | Our model · unseen test sets | median SROCC | Mixed |
| add KoNViD-1k to training, test on unseen full datasets (ablation) | Our model · CVD2014 (full), LIVE-Qualcomm (full), LIVE-VQC (full) | median SROCC: 0.3390 +0.2579 (CVD2014 train, LIVE-Qualcomm test) | Improved |
| increase proportion of training data (ablation) | MDTVSFA · Four datasets | mean SROCC | Improved |
Closest papers by meaning
- PTM-VQA: Efficient Video Quality Assessment Leveraging Diverse PreTrained Models from the Wild (CVPR 2024 · similarity 0.88)
- AIGV-Assessor: Benchmarking and Evaluating the Perceptual Quality of Text-to-Video Generation with LMM (CVPR 2025 · similarity 0.87)
- Scaling and Masking: A New Paradigm of Data Sampling for Image and Video Quality Assessment (AAAI 2024 · similarity 0.87)
- Knowledge Guided Semi-supervised Learning for Quality Assessment of User Generated Videos (AAAI 2024 · similarity 0.87)
- KVQ: Boosting Video Quality Assessment via Saliency-guided Local Perception (CVPR 2025 · similarity 0.86)
- Modular Blind Video Quality Assessment (CVPR 2024 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
