# A Deep Learning based No-reference Quality Assessment Model for UGC Videos
###### tags: `paper` `vqa` `2023`

## Paper Information
- Publish time: Oc. 2022 (ACM MM)
- Authors: Shanghai Jiao Tong University (SJTU)
- Github: https://github.com/sunwei925/SimpleVQA
- 2023 NTIRE VQA baseline

## OSS
#### Problem
- No-reference quality assessment for UGC videos
#### Background
- Previous studies extract frame-level features rom all video frames and have a very high computational complexity, making them difficult to apply to real-world scenarios.
#### Solution ideas
- The model utilizes very sparse frames to extract spatial features and dense frames (i.e. the video chunk) with a very low spatial resolution to extract motion features, which thereby has low computational complexity.
#### Result
- The proposed model achieves the best performance on five popular UGC VQA databases.
#### Insight
- The model is simple and straightforward, so it would be more possible for real-world usage.

## Personal Notes
- In order to ensure the Quality of Experience (QoE) of end-users, the service providers need to monitor the quality of UGC videos in the entire streaming media link, including but not limited to video uploading, compressing, post-processing, transmitting, etc.
- Model