On the use of deep learning for blind image quality assessment

By Simone Bianco, Luigi Celona, Paolo Napoletano, Raimondo Schettini. Signal, Image and Video Processing, 2016

365 citations recorded by the source corpus.

In this work, we investigate the use of deep learning for distortion-generic blind image quality assessment. We report on different design choices, ranging from the use of features extracted from pre-trained convolutional neural networks (CNNs) as a generic image description, to the use of features extracted from a CNN fine-tuned for the image quality task. Our best proposal, named DeepBIQ, estimates the image quality by average-pooling the scores predicted on multiple subregions of the original image. Experimental results on the LIVE In the Wild Image Quality Challenge Database show that DeepBIQ outperforms the state-of-the-art methods compared, having a linear correlation coefficient with human subjective scores of almost 0.91. These results are further confirmed also on four benchmark databases of synthetically distorted images: LIVE, CSIQ, TID2008, and TID2013.

Subfields: Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/d45e2cddbc5e8890fa56e343285f01a23dc5efe9

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

Citations in the corpus

MetaIQA: Deep Meta-Learning for No-Reference Image Quality Assessment (2020, methodology): “TID2013 [38] 25 3,000 24 [0, 9] KADID-10K [27] 81 10,125 25 [1, 5]”
No-Reference Image Quality Assessment via Transformers, Relative Ranking, and Self-Consistency (2021, background): “By success of deep learning [29, 30] in many computer vision tasks, different approaches utilize deep learning for NR-IQA [31, 32, 33, 34, 35, 36, 37, 38, 39, 36, 40, 41, 42, 43, 10, 44].”
End-to-End Blind Image Quality Assessment Using Deep Neural Networks (2018, methodology): “Nevertheless, their proposal is not end-to-end optimized and involves heavy manual parameter adjustments [13].”
From Patches to Pictures (PaQ-2-PiQ): Mapping the Perceptual Space of Picture Quality (2019, background): “The authors of [51], the code of which is not made available, reported high results, but we have been unable to reproduce their numbers, even with more efficient networks.”
RankIQA: Learning from Rankings for No-Reference Image Quality Assessment (2017, methodology): “Research has mostly focussed on the more realist scenario of NR-IQA where the image quality of an image without any reference image has to be estimated.”
MANIQA: Multi-dimension Attention Network for No-Reference Image Quality Assessment (2022, methodology): “With the success of deep learning and Convolution Neural Network (CNN) in many computer vision tasks, CNN-based methods have significantly outperformed previous approaches in terms of real-world distortions by directly extracting discriminative features from LQ images [1,3,14,20,26,37,39,45,59,61].”
NIMA: Neural Image Assessment (2017, methodology): “[7] feed multiple image crops to a deep CNN, whereas our method takes only the rescaled image.”
Image Quality Assessment by Comparing CNN Features between Images (2016, background): “Many different approaches for measuring image quality have been proposed, including structural similarity [51, 56], color difference [15], spatial extensions of color difference formulae [22, 47, 54, 58], simulation of detail visibility [38, 34], scene statistics [44], low- and mid-level visual prop”
Real-World Underwater Enhancement: Challenges, Benchmarks, and Solutions Under Natural Light (2019, background): “Recently, data-driven CNNs comprising the information from human labels for non-reference quality assessment of natural images have rapidly developed and achieved remarkable performance [51], [52], [53].”
KonIQ-10k: An Ecologically Valid Database for Deep Learning of Blind Image Quality Assessment (2019, methodology): “Recently, deep learning has achieved promising results in a number of computer vision tasks: image classiﬁcation [9] [10], object detection [11] [12], and BIQA [13] [14] [15].”