Article
Computer Science, Information Systems
Tomas Mizdos, Marcus Barkowsky, Miroslav Uhrina, Peter Pocta
Summary: Researchers are developing new methods to predict media quality by mapping technical parameters to perceived quality using deep learning algorithms. This study aims to extract more training data from previous image datasets, evaluate the connection between different distortion types, and verify the effectiveness of the method through subjective experiments.
MULTIMEDIA TOOLS AND APPLICATIONS
(2021)
Article
Chemistry, Multidisciplinary
Claudio Navar Valdebenito Maturana, Ana Lucila Sandoval Orozco, Luis Javier Garcia Villalba
Summary: This article discusses the significance of using metrics to evaluate synthetic images generated by generative adversarial networks. By analyzing metrics and datasets, researchers can understand and improve the performance and quality of generative adversarial networks.
APPLIED SCIENCES-BASEL
(2023)
Article
Computer Science, Information Systems
Kyohoon Sim, Jiachen Yang, Wen Lu, Xinbo Gao
Summary: This paper addresses the correlation between stereoscopic image quality assessment (SIQA) and semantic recognition and proposes a new method that utilizes deep convolutional neural networks to extract binocular semantic features, solving the limitation of dataset size and achieving good prediction results on benchmark datasets.
IEEE TRANSACTIONS ON MULTIMEDIA
(2022)
Review
Engineering, Biomedical
Lucie Leveque, Meriem Outtas, Hantao Liu, Lu Zhang
Summary: Healthcare professionals increasingly rely on medical images and videos in their daily clinical practice, with the quality of visual content being a crucial factor in patient care. However, such content is not self-explanatory and may be affected by degradations and artifacts, impacting both viewer experience and clinical practice. Recent literature has focused on understanding how medical experts perceive visual content quality through subjective assessments.
PHYSICS IN MEDICINE AND BIOLOGY
(2021)
Article
Computer Science, Information Systems
Wenxin Yu, Xuewen Zhang, Yunye Zhang, Zhiqiang Zhang, Jinjia Zhou
Summary: The paper introduces a method to evaluate the quality of single generated images in text-to-image synthesis, which includes designing a specific dataset and proposing a learning model. Experiments show that the method can partially solve this problem.
Article
Computer Science, Artificial Intelligence
Jupo Ma, Jinjian Wu, Leida Li, Weisheng Dong, Xuemei Xie, Guangming Shi, Weisi Lin
Summary: This paper proposes a novel blind image quality assessment method based on active inference, which mimics the working mechanism of the human visual system for image quality evaluation. The method achieves competitive performance in predicting primary content and measuring comprehensive quality degradation.
IEEE TRANSACTIONS ON IMAGE PROCESSING
(2021)
Article
Engineering, Electrical & Electronic
Zhaoqing Pan, Hao Zhang, Jianjun Lei, Yuming Fang, Xiao Shao, Nam Ling, Sam Kwong
Summary: Deep neural networks have achieved great performance on blind Image Quality Assessment (IQA). In this paper, a Distortion-Aware Convolutional Neural Network (DACNN) is proposed, which works effectively for both synthetically distorted images and authentically distorted images.
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY
(2022)
Review
Computer Science, Artificial Intelligence
Dhruv Sharma, Chhavi Dhiman, Dinesh Kumar
Summary: Automatic Visual Captioning (AVC) generates syntactically and semantically correct sentences by describing important objects, attributes, and their relationships with each other. It is widely used in various applications such as assistance for the visually impaired, human-robot interaction, video surveillance systems, scene understanding, etc. With the unprecedented success of deep-learning in Computer Vision and Natural Language Processing, the past few years have seen a surge of research in this domain.
EXPERT SYSTEMS WITH APPLICATIONS
(2023)
Article
Computer Science, Information Systems
Nandhini Chockalingam, Brindha Murugan
Summary: The article introduces the methods of dense convolution network (DSC-Net) and multimodal dense convolution network (MDSC-Net) for image quality assessment. DSC-Net improves the quality of image representation by reducing the number of parameters and addressing the issue of overfitting. MDSC-Net combines texture features and spatial features to enhance the performance of image quality prediction through multimodal data.
FRONTIERS OF INFORMATION TECHNOLOGY & ELECTRONIC ENGINEERING
(2023)
Article
Construction & Building Technology
Yuwei Wang, Dorukalp Durmus
Summary: Adaptive lighting systems can adjust light output based on the characteristics of indoor environments to improve visual comfort and performance. This study investigates the relationship between perceived quality of indoor environments, personality, and computational image quality metrics. The results show a strong correlation between perceived colorfulness and clarity and complexity, while personality traits do not impact subjective evaluations of indoor environmental images.
Article
Computer Science, Information Systems
Pengfei Guo, Lang He, Shuangyin Liu, Delu Zeng, Hantao Liu
Summary: This paper investigates the performance of five popular enhancement algorithms for underwater images and analyzes their impact on perceptual quality. It also evaluates the visual quality objectively, aiming to develop objective metrics for automatic assessment of underwater image enhancement quality.
IEEE TRANSACTIONS ON MULTIMEDIA
(2022)
Article
Computer Science, Information Systems
Leonardo Galteri, Lorenzo Seidenari, Pietro Bongini, Marco Bertini, Alberto Del Bimbo
Summary: This article discusses the issue of image quality assessment. Deep networks are often used to evaluate image quality based on human-provided scores, but this approach may not generalize well to unseen distortions. To address the time-consuming nature of human-based assessment, the article proposes the use of language generation tasks to evaluate the quality of restored images.
ACM TRANSACTIONS ON MULTIMEDIA COMPUTING COMMUNICATIONS AND APPLICATIONS
(2022)
Article
Engineering, Electrical & Electronic
Xuting Lan, Mingliang Zhou, Xueyong Xu, Xuekai Wei, Xingran Liao, Huayan Pu, Jun Luo, Tao Xiang, Bin Fang, Zhaowei Shang
Summary: In this paper, a framework based on two feature extraction networks and a multilevel feature fusion (MFF) network is proposed. This method can obtain multilevel degradation features and capture local and global feature information contained in these features for the prediction of distorted images. Experimental results show that the proposed method achieves greatly improved prediction accuracy and performance on five standard databases.
IEEE TRANSACTIONS ON BROADCASTING
(2023)
Article
Computer Science, Artificial Intelligence
S. K. Mahaboob Basha, S. A. Kalaiselvan
Summary: This paper discusses the importance of QoE predictions in multimedia applications and proposes techniques to improve accuracy and generalization. It suggests using more complex neural network architectures and diverse datasets to enhance QoE predictions, leading to improved user satisfaction and adoption.
JOURNAL OF INTELLIGENT & FUZZY SYSTEMS
(2023)
Article
Computer Science, Artificial Intelligence
Yuan Chen, Yang Zhao, Li Cao, Wei Jia, Xiaoping Liu
Summary: This paper proposes a blind cartoon image quality assessment method based on convolutional neural networks, and improves the network's robustness by establishing a large-scale dataset and implementing a random degradation strategy. Experimental results demonstrate the effectiveness and robustness of the proposed method on synthetic and real-world cartoon image datasets.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2023)
Article
Computer Science, Artificial Intelligence
Jaehun Kim, Julian Urbano, Cynthia C. S. Liem, Alan Hanjalic
NEURAL COMPUTING & APPLICATIONS
(2020)
Article
Computer Science, Artificial Intelligence
Jingkuan Song, Tao He, Lianli Gao, Xing Xu, Alan Hanjalic, Heng Tao Shen
INTERNATIONAL JOURNAL OF COMPUTER VISION
(2020)
Article
Chemistry, Analytical
Tianyi Zhang, Abdallah El Ali, Chen Wang, Alan Hanjalic, Pablo Cesar
Summary: The study proposes a correlation-based emotion recognition algorithm (CorrNet) to recognize the valence and arousal of each instance using wearable physiological signals, achieving promising recognition accuracies in indoor and outdoor environments. Results show that instance segment lengths between 1-4s result in the highest recognition accuracies, and large amounts of neutral V-A labels affect the recognition performance. The study also found that the accuracies between laboratory-grade and wearable sensors are comparable, even under low sampling rates.
Article
Computer Science, Information Systems
Elvin Isufi, Matteo Pocchiari, Alan Hanjalic
Summary: The study introduces a joint graph convolutional model that balances accuracy and diversity in recommender systems by learning convolutions from nearest neighbor and furthest neighbor graphs, with the information between the two modules balanced in training through a regularizer inspired by multi-kernel learning. The proposed method can significantly improve catalog coverage or diversity within the list, with diversity gains up to seven times by trading as little as 1% in accuracy.
INFORMATION PROCESSING & MANAGEMENT
(2021)
Article
Computer Science, Artificial Intelligence
Xing Xu, Tan Wang, Yang Yang, Alan Hanjalic, Heng Tao Shen
Summary: This article introduces an innovative answer-centric approach called radial graph convolutional network (Radial-GCN) for visual question generation (VQG). Experimental results demonstrate the superiority of this method over reference methods on three benchmark datasets, and even boost the performance of state-of-the-art VQA methods significantly in the challenging zero-shot VQA task.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2021)
Article
Computer Science, Information Systems
Xing Xu, Yifan Wang, Yixuan He, Yang Yang, Alan Hanjalic, Heng Tao Shen
Summary: In this study, a novel CMHF framework is proposed for directly learning the image-sentence similarity by fusing multimodal features with inter- and intra-modality relations incorporated. The framework utilizes flexible attention mechanisms to generate effective attention flows within and across the modalities of images and sentences, capturing high-level interactions between visual regions in images and words in sentences. The structured objective with ranking loss constraint in CMHF is demonstrated to effectively learn the image-sentence similarity based on the fused fine-grained features of different modalities, achieving state-of-the-art matching performance.
ACM TRANSACTIONS ON MULTIMEDIA COMPUTING COMMUNICATIONS AND APPLICATIONS
(2021)
Article
Computer Science, Information Systems
Manel Slokom, Alan Hanjalic, Martha Larson
Summary: This paper introduces a new privacy solution called PerBlur for protecting user privacy while training a recommender system, by adding and removing items from user profiles to generate obfuscated user-item matrix. Results show that gender obfuscation impacts the fairness and diversity of recommender system results, highlighting the importance of maintaining fairness and enhancing diversity for user recommendations.
INFORMATION PROCESSING & MANAGEMENT
(2021)
Article
Computer Science, Artificial Intelligence
Xing Xu, Kaiyi Lin, Yang Yang, Alan Hanjalic, Heng Tao Shen
Summary: This article proposes a novel method called Joint Feature Synthesis and Embedding (JFSE), which utilizes two coupled conditional Wassertein GAN modules to synthesize meaningful and correlated multimodal features, and employs advanced distribution alignment schemes and cycle-consistency constraints to preserve semantic compatibility and enable knowledge transfer in a shared embedding space. Experimental results show that the JFSE method achieves significant accuracy improvement in standard retrieval and newly explored zero-shot and generalized zero-shot retrieval tasks.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2022)
Article
Engineering, Multidisciplinary
Li Zou, Xiu-Xiu Zhan, Jie Sun, Alan Hanjalic, Huijuan Wang
Summary: This study focuses on predicting temporal networks using interpretable learning algorithms like Lasso Regression and Random Forest. The results show that the next step activity of a particular link is mainly influenced by its current activity and links strongly correlated in the time series and close in distance in the aggregated network.
IEEE TRANSACTIONS ON NETWORK SCIENCE AND ENGINEERING
(2022)
Article
Engineering, Electrical & Electronic
Bishwadeep Das, Alan Hanjalic, Elvin Isufi
Summary: This paper discusses the importance of connectivity information for data processing on expanding graphs with new nodes. By modeling the attachment of new nodes without connectivity information and showing implicit constraints on spectral perturbation, the paper provides a task-driven data processing approach. Numerical results confirm the superior performance of the proposed approach in the absence of connectivity information.
IEEE TRANSACTIONS ON SIGNAL AND INFORMATION PROCESSING OVER NETWORKS
(2022)
Article
Computer Science, Theory & Methods
Omar F. Robledo, Xiu-Xiu Zhan, Alan Hanjalic, Huijuan Wang
Summary: This paper investigates the impact of network topology on the performance of network embedding algorithms in link prediction. The results show that a higher clustering coefficient leads to better link prediction performance, except for Matrix Factorisation which is not sensitive to changes in clustering coefficient. The study found that the algorithms tend to assign a higher likelihood of connection to node pairs with a higher number of common neighbors, regardless of the clustering coefficient. The predicted networks have more triangles and higher clustering coefficient as a result. The findings suggest that increasing the clustering coefficient improves link prediction performance, except for Matrix Factorisation.
APPLIED NETWORK SCIENCE
(2022)
Proceedings Paper
Computer Science, Artificial Intelligence
Roger Zhe Li, Julian Urbano, Alan Hanjalic
Summary: This paper presents a method to address mainstream bias by adding an autoencoder layer, improving recommendations for nonmainstream users.
WSDM '21: PROCEEDINGS OF THE 14TH ACM INTERNATIONAL CONFERENCE ON WEB SEARCH AND DATA MINING
(2021)
Proceedings Paper
Computer Science, Information Systems
Roger Zhe Li, Julian Urbano, Alan Hanjalic
Summary: Direct optimization of IR metrics in ranking-based recommender systems may not necessarily lead to the best performance, as shown in an experimental study comparing the relative merits of different IR metrics. RBP-inspired losses offer consistent and clear benefits, especially for more active users, challenging the current research practice of optimizing and evaluating the same metric in recommendation systems.
SIGIR '21 - PROCEEDINGS OF THE 44TH INTERNATIONAL ACM SIGIR CONFERENCE ON RESEARCH AND DEVELOPMENT IN INFORMATION RETRIEVAL
(2021)
Article
Acoustics
Xinsheng Wang, Tingting Qiao, Jihua Zhu, Alan Hanjalic, Odette Scharenborg
Summary: This paper introduces a new speech technology task - speech-to-image generation framework, showcasing its potential applications in unwritten languages. Through experiments, the effectiveness of S2IGAN in synthesizing high-quality and semantically-consistent images has been demonstrated.
IEEE-ACM TRANSACTIONS ON AUDIO SPEECH AND LANGUAGE PROCESSING
(2021)
Article
Mathematics, Interdisciplinary Applications
Jaehun Kim, Julian Urbano, Cynthia C. S. Liem, Alan Hanjalic
FRONTIERS IN APPLIED MATHEMATICS AND STATISTICS
(2019)
Article
Engineering, Electrical & Electronic
Xueyu Han, Ishtiaq Rasool Khan, Susanto Rahardja
Summary: This paper proposes a clustering-based TMO method by embedding human visual system models to adapt to different HDR scenes. The method reduces computational complexity using a hierarchical scheme for clustering and enhances local contrast by superimposing details and controlling color saturation by limiting the adaptive saturation parameter. Experimental results show that the proposed method achieves improvements in generating high quality tone-mapped images compared to competing methods.
SIGNAL PROCESSING-IMAGE COMMUNICATION
(2024)
Article
Engineering, Electrical & Electronic
Zuopeng Zhao, Tianci Zheng, Kai Hao, Junjie Xu, Shuya Cui, Xiaofeng Liu, Guangming Zhao, Jie Zhou, Chen He
Summary: The research team developed a handheld phone detection network called YOLO-PAI, which successfully achieved real-time detection and underwent testing under various conditions. Experimental results show that YOLO-PAI reduces network structure parameters and computational costs while maintaining accuracy, outperforming other popular networks in terms of speed and accuracy.
SIGNAL PROCESSING-IMAGE COMMUNICATION
(2024)
Article
Engineering, Electrical & Electronic
Vivek Sharma, Ashish Kumar Tripathi, Purva Daga, M. Nidhi, Himanshu Mittal
Summary: In this study, a novel ClGan method is proposed for automated plant disease detection. The method reduces the number of parameters and addresses the issues of vanishing gradients, training instability, and non-convergence by using an encoder-decoder network. Additionally, an improved loss function is introduced to stabilize the learning process and optimize weights effectively. Furthermore, a new plant leaf classification method called ClGanNet is introduced, achieving 99.97% training accuracy and 99.04% testing accuracy using the least number of parameters.
SIGNAL PROCESSING-IMAGE COMMUNICATION
(2024)
Article
Engineering, Electrical & Electronic
Seongeun Kim, Chang-Ock Lee
Summary: This article introduces a method for segmenting individual teeth in human teeth images by using deep neural networks to obtain pseudo edge-regions and applying active contour models for segmentation.
SIGNAL PROCESSING-IMAGE COMMUNICATION
(2024)