Semantic Information Oriented No-Reference Video Quality Assessment

Description

SIONR code for the following papers:

Wei Wu, Qinyao Li, Zhenzhong Chen, Shan Liu. Semantic Information Oriented No-Reference Video Quality Assessment.IEEE SPL Paper

Feature Extraction

python generate_CNNfeatures.py

Test Demo

The model weights provided in model/SIONR.pt are the saved weights when running a random split of KoNViD-1k. The random split is shown in data/train_val_test_split.xlsx, which contains video file names, scores, and train/validation/test split assignment (random).

python test_demo.py

The test results are shown in result/test_result.xlsx.

The mean (std) values of the first twenty index splits (60%:20%:20% train:val:test)

	KoNViD-1k	LIVE-VQC
PLCC	0.8180 (0.0172)	0.7821 (0.0355)
SROCC	0.8109 (0.0200)	0.7361 (0.0446)
RMSE	0.3688 (0.0160)	10.4744 (0.6052)

UGC-VQA Datasets

BVQA Dataset	Download	Paper
KoNViD-1k (2017)	KoNViD-1k	Hosu et al. QoMEX'17
LIVE-VQC (2018)	LIVE-VQC	Sinno et al. TIP'19

NR-VQA models

Model	Download	Paper
NIQE	NIQE	Mittal et al. IEEE SPL'12
BRISQUE	BRISQUE	Mittal et al. TIP'12
CORNIA	CORNIA	Ye et al. CVPR'12
V-BLIINDS	V-BLIINDS	Saad et al. TIP'13
HIGRADE	HIGRADE	Kundu et al. TIP'17
FRIQUEE	FRIQUEE	Ghadiyaram et al. JOV'17
VSFA	VSFA	Li et al. ACM MM'19
TLVQM	nr-vqa-consumervideo	Korhenen et al. TIP'19

Citation

@article{wu2021semantic,
  title={Semantic Information Oriented No-Reference Video Quality Assessment},
  author={Wu, Wei and Li, Qinyao and Chen, Zhenzhong and Liu, Shan},
  journal={IEEE Signal Processing Letters}, 
  year={2021}
}

Requirement

PyTorch, scikit-video, pandas

Contact

wuwei_whu AT whu DOT edu DOT cn.

renyuzhuo/SIONR