Summary of Video Quality Assessment Based on Swin Transformerv2 and Coarse to Fine Strategy, by Zihao Yu et al.
Video Quality Assessment Based on Swin TransformerV2 and Coarse to Fine Strategy
by Zihao Yu, Fengbin Guan, Yiting Lu, Xin Li, Zhibo Chen
First submitted to arxiv on: 16 Jan 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Machine Learning (cs.LG); Image and Video Processing (eess.IV)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The proposed non-reference video quality assessment method tackles the challenge of evaluating distorted video without access to high-definition references. By combining a pre-trained spatial perception module, a lightweight temporal fusion module, and a transformer-based architecture, this approach effectively addresses the NR-VQA task. The model’s Swin Transformer V2 component extracts local-level spatial features, which are then fused through transformer layers for multi-stage representation. A temporal transformer is also utilized to merge spatiotemporal features across videos of varying bitrates. To further enhance performance, a coarse-to-fine contrastive strategy is incorporated to distinguish features from compressed videos. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This study focuses on assessing the quality of distorted video without a reference high-definition version. Researchers developed a new method that combines spatial and temporal processing. They trained a model using multiple image datasets and then applied it to videos with different qualities. The goal was to develop an accurate way to evaluate video quality without knowing what the original looks like. |
Keywords
* Artificial intelligence * Spatiotemporal * Transformer
