A Visual Saliency-Based Neural Network Architecture for No-Reference Image Quality Assessment

Ryu, Jihyoung

doi:10.3390/app12199567

Cited by 7 publications

(5 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In [36], the authors improved this approach by applying a nonlinear bilateral smoothing filtering technique and a nearest neighbor sampling approach [37]. Prior to deep feature extraction, Ryu [38] introduced a static saliency detection module first to identify those regions which humans tend to pay more attention to. In contrast, Celona and Schettini [39] devised a deep architecture which handles images at different scales.…”

Section: Literature Reviewmentioning

confidence: 99%

No-Reference Image Quality Assessment Using the Statistics of Global and Local Image Features

Varga

2023

Electronics

View full text Add to dashboard Cite

Methods of image quality assessment are widely used for ranking computer vision algorithms or controlling the perceptual quality of video and streaming applications. The ever-increasing number of digital images has encouraged the research in this field at an accelerated pace in recent decades. After the appearance of convolutional neural networks, many researchers have paid attention to different deep architectures to devise no-reference image quality assessment algorithms. However, many systems still rely on handcrafted features to ensure interpretability and restrict the consumption of resources. In this study, our efforts are focused on creating a quality-aware feature vector containing information about both global and local image features. Specifically, the research results of visual physiology indicate that the human visual system first quickly and automatically creates a global perception before gradually focusing on certain local areas to judge the quality of an image. Specifically, a broad spectrum of statistics extracted from global and local image features is utilized to represent the quality-aware aspects of a digital image from various points of view. The experimental results demonstrate that our method’s predicted quality ratings relate strongly with the subjective quality ratings. In particular, the introduced algorithm was compared with 16 other well-known advanced methods and outperformed them by a large margin on 9 accepted benchmark datasets in the literature: CLIVE, KonIQ-10k, SPAQ, BIQ2021, TID2008, TID2013, MDID, KADID-10k, and GFIQA-20k, which are considered de facto standards and generally accepted in image quality assessment.

show abstract

Section: Literature Reviewmentioning

confidence: 99%

No-Reference Image Quality Assessment Using the Statistics of Global and Local Image Features

Varga

2023

Electronics

View full text Add to dashboard Cite

show abstract

“…In InceptionResnetV2, input images of size 299 × 299 pixels (Ryu, 2023) undergo convolutional layers to extract features (Wan et al, 2019). Inception modules capture multi-scale features with different filter sizes (Nazir et al, 2019), followed by a filter expansion layer (1 × 1 convolution without activation) to match input depth (Ryu, 2022). Output feature maps are merged through filter concatenation.…”

Section: Dataset Classification Mechanism With Inceptionresnetv2mentioning

confidence: 99%

Enzyme dosage detection to degrade feathers in edible bird’s nests: A comparative convolutional neural networks study

Liana,

Arifiandika,

Rohmatulloh

et al. 2023

AFSSAAE

View full text Add to dashboard Cite

Edible Bird’s Nest (EBN), a costly food product made from swiftlet’s saliva, has encountered a longstanding problem of plucking the swiftlet’s feather from the nests. The destructive and inefficient manual process of plucking the feathers can be substituted with a serine protease enzyme alternative. Accurate detection of enzyme dosage is crucial for ensuring efficient feather degradation with cost-effective enzyme usage. This research employed the transfer learning method using pretrained Convolutional Neural Networks (Pt-CNN) to detect enzyme dosage based on EBN’s images. This study aimed to compare the image classification mechanisms, architectures, and performance of three Pt-CNN: Resnet50, InceptionResnetV2, and EfficientNetV2S. InceptionResnetV2, using parallel convolutions and residual networks, significantly contributes to learning rich informative features. Consequently, the InceptionResnetV2 model achieved the highest accuracy of 96.18%, while Resnet50 and EfficientNetV2S attained only 30.44% and 17.82%, respectively. The differences in architecture complexity, parameter count, dataset characteristics, and image resolution also play a role in the performance disparities among the models. The study’s findings aid future researchers in streamlining model selection when facing limited datasets by understanding the reasons for the model’s performance and contributing to a non-destructive and quick solution for EBN’s cleaning process.

show abstract

“…IQA is crucial to make sure that the image is free from distortions like noise or blur so that the objects in the image can be identi ed clearly which will be helpful for surveillance. IQA can be categorized into two forms and they are SIQA and OIQA [2].The application of drones is becoming highly popular nowadays ranging from agriculture to military for surveillance. In agriculture, the application of drones helps many farmers to monitor crop productivity and reduces the overall farming amount [11,12].…”

Section: Introductionmentioning

confidence: 99%

Object Detection in Enhanced Drone Imagery using Extended ESRGAN with SSD

Subiksha,

Pattabiraman,

Jaincy

2023

Preprint

View full text Add to dashboard Cite

Image enhancement and detection have achieved great attention in the field of deep learning. Image quality tops the list to the viewers because if the quality of an image is not satisfied by the consumer then the other features like cost, and warranty becomes less important. IQA is crucial to make sure that the image is free from distortions so that the objects in the image can be identified clearly. This study aims to predict objects in images captured by drones. But these images contain some degradation as it is captured at flight altitudes. An Extended ESRGAN algorithm has been proposed for enhancement. The spatial resolution of the image is stabilized using a Bi-linear interpolation algorithm. Furthermore, perceptual metrics which include SSIM, RMSE, and PSNR are computed to show that the proposed methodology gives better results in terms of image quality. After obtaining a high-resolution image from a low-resolution image using a Super resolution algorithm, Objects in the processed image are identified using DL and OpenCV. Single shot Detector has been utilized to detect the objects so that it will be helpful for the military to track the movements of enemies. This process has been implemented for normal images as well to prove that the proposed system gives good results in terms of enhancement and detection. In addition, a comparison between existing models and the proposed model has been performed. Results showed that the accuracy score of object prediction using SSD was 79% whereas with YOLOv5 it was only 68%.

show abstract

A Visual Saliency-Based Neural Network Architecture for No-Reference Image Quality Assessment

Cited by 7 publications

References 37 publications

No-Reference Image Quality Assessment Using the Statistics of Global and Local Image Features

No-Reference Image Quality Assessment Using the Statistics of Global and Local Image Features

Enzyme dosage detection to degrade feathers in edible bird’s nests: A comparative convolutional neural networks study

Object Detection in Enhanced Drone Imagery using Extended ESRGAN with SSD

Contact Info

Product

Resources

About