2021
DOI: 10.48550/arxiv.2101.10955
|View full text |Cite
Preprint
|
Sign up to set email alerts
|

RAPIQUE: Rapid and Accurate Video Quality Prediction of User Generated Content

Zhengzhong Tu,
Xiangxu Yu,
Yilin Wang
et al.

Abstract: Blind or no-reference video quality assessment of user-generated content (UGC) has become a trending, challenging, unsolved problem. Accurate and efficient video quality predictors suitable for this content are thus in great demand to achieve more intelligent analysis and processing of UGC videos. Previous studies have shown that natural scene statistics and deep learning features are both sufficient to capture spatial distortions, which contribute to a significant aspect of UGC video quality issues. However, … Show more

Help me understand this report
View published versions

Search citation statements

Order By: Relevance

Paper Sections

Select...
2
1
1

Citation Types

0
15
0

Year Published

2021
2021
2022
2022

Publication Types

Select...
7

Relationship

2
5

Authors

Journals

citations
Cited by 9 publications
(15 citation statements)
references
References 66 publications
0
15
0
Order By: Relevance
“…V-MEON [25] merges feature extraction and regression into a single stage where both are jointly optimized within a multitask DNN. Among hybrid models, RAPIQUE [12] efficiently combines low-level NSS features with high-level deep learning features which are used to train a single regressor head that predicts video quality scores. PVQ [23] extracts both spatial and temporal features using RoIPool (region-of interest pooling) and SoIPool (segment-of-interest pooling) layers, using them to learn local to global spatio-temporal quality relationships.…”
Section: A Subjective Video Quality Assessmentmentioning
confidence: 99%
See 3 more Smart Citations
“…V-MEON [25] merges feature extraction and regression into a single stage where both are jointly optimized within a multitask DNN. Among hybrid models, RAPIQUE [12] efficiently combines low-level NSS features with high-level deep learning features which are used to train a single regressor head that predicts video quality scores. PVQ [23] extracts both spatial and temporal features using RoIPool (region-of interest pooling) and SoIPool (segment-of-interest pooling) layers, using them to learn local to global spatio-temporal quality relationships.…”
Section: A Subjective Video Quality Assessmentmentioning
confidence: 99%
“…This simple yet effective observation has led to a wide variety of IQA/VQA models that map measured statistical model parameters to perceptual quality in different bandpass domains [6]- [8], [10], [11]. One very recent model called RAPIQUE [12] adopts a particularly efficient strategy, whereby instead of hand-crafting features in a variety of bandpass spaces, it deploys an NSS-based feature extractor which is applied on each feature transform. RAPIQUE's powerful and compute-efficient strategy attains SOTA performance across a wide variety of subjective databases.…”
Section: A Spatial Feature Designmentioning
confidence: 99%
See 2 more Smart Citations
“…Wang et al[9] create a large scale UGC video dataset and propose a DNN-based framework to thoroughly analyze importance of content, technical quality and compression level in perceptual quality. Tu et al[7] proposed an efficient model for predicting the subjective quality of UGC videos…”
mentioning
confidence: 99%