A5002277899- Advanced Data Compression Techniques
- Video Coding and Compression Technologies
- Image and Video Quality Assessment
- Advanced Vision and Imaging
- Advanced Image and Video Retrieval Techniques
- Cooperative Communication and Network Coding
- Image and Signal Denoising Methods
- Advanced Image Processing Techniques
- Advanced Wireless Network Optimization
- Advanced MIMO Systems Optimization
- Visual Attention and Saliency Detection
- Wireless Communication Security Techniques
- Plant Physiology and Cultivation Studies
- Advanced Wireless Communication Techniques
- Multimedia Communication and Technology
- Error Correcting Code Techniques
- Image Enhancement Techniques
- Video Analysis and Summarization
- Image Retrieval and Classification Techniques
- Multimodal Machine Learning Applications
- Wireless Networks and Protocols
- Sparse and Compressive Sensing Techniques
- Medical Image Segmentation Techniques
- Human Pose and Action Recognition
- Mobile Ad Hoc Networks
Hong Kong Polytechnic University
2021-2025
University at Buffalo, State University of New York
2014-2024
Peng Cheng Laboratory
2019-2024
Southwest University
2024
Zhengzhou Fruit Research Institute
2014-2023
Chinese Academy of Agricultural Sciences
2016-2023
Southwest University of Science and Technology
2023
Ministry of Agriculture and Rural Affairs
2023
National Cheng Kung University Hospital
1998-2021
Huazhong Agricultural University
2021
This paper presents the first complete design to apply compressive sampling theory sensor data gathering for large-scale wireless networks. The successful scheme developed in this research is expected offer fresh frame of mind both applications and We consider scenario which a large number nodes are densely deployed readings spatially correlated. proposed able reduce global scale communication cost without introducing intensive computation or complicated transmission control. load balancing...
Traditional full-reference image quality assessment (IQA) metrics generally predict the of distorted by measuring its deviation from a perfect called reference image. When is not fully available, reduced-reference and no-reference IQA may still be able to derive some characteristics images, then measure image's these characteristics. In this paper, contrary conventional metrics, we utilize new "reference" pseudo-reference (PRI) PRI-based blind (BIQA) framework. Different traditional image,...
The general purpose of seeing a picture is to attain information as much possible. With it, we in this paper devise new no-reference/blind metric for image quality assessment (IQA) contrast distortion. For local details, lirst roughly remove predicted regions an since unpredicted remains are information. We then compute entropy particular areas maximum via visual saliency. From global perspective, compare the histogram with uniformly distributed symmetric Kullback-Leibler divergence....
In this paper, we investigate the problem of image contrast enhancement. Most existing relevant technologies often suffer from drawback excessive enhancement, thereby introducing noise/artifacts and changing visual attention regions. One frequently used solution is manual parameter tuning, which is, however, impractical for most applications since it labor intensive time consuming. research, find that saliency preservation can help produce appropriately enhanced images, i.e., improved...
Deep convolutional neural networks (CNN) have recently been shown to generate promising results for aesthetics assessment. However, the performance of these deep CNN methods is often compromised by constraint that network only takes fixed-size input. To accommodate this requirement, input images need be transformed via cropping, warping, or padding, which alter image composition, reduce resolution, cause distortion. Thus original impaired because potential loss fine grained details and...
Human selection has a long history of transforming crop genomes. Peach (Prunus persica) undergone more than 5000 years domestication that led to remarkable changes in series agronomically important traits, but genetic bases underlying these and the effects artificial on genomic diversity are not well understood.Here, we report comprehensive analysis peach evolution based genome sequences 480 wild cultivated accessions. By focusing set quantitative trait loci (QTLs), provide evidence...
Finding relevant moments and highlights in videos according to natural language queries is a highly valuable common need the current video content explosion era. Nevertheless, jointly conducting moment retrieval highlight detection an emerging research topic, even though its component problems some related tasks have already been studied for while. In this paper, we present first unified framework, named Unified Multi-modal Transformers (UMT), capable of realizing such joint optimization...
Multi-instance learning (MIL) is an effective paradigm for whole-slide pathological images (WSIs) classification to handle the gigapixel resolution and slide-level label. Prevailing MIL methods primarily focus on improving feature extractor aggregator. However, one deficiency of these that bag contextual prior may trick model into capturing spurious correlations between bags labels. This a confounder limits performance existing methods. In this paper, we propose novel scheme, Interventional...
3D human pose estimation has been researched for decades with promising fruits. lifting is one of the research directions toward task where both estimated and ground truth data are used training. Existing works mainly focus on improving performance pose, but they usually underperform when testing data. We observe that can be easily improved by preparing good quality 2D such as fine-tuning or using advanced detectors. As such, we concentrate via future improvement more Towards this goal, a...
The task of instance segmentation in remote sensing images, aiming at performing per-pixel labeling objects the level, is great importance for various civil applications. Despite previous successes, most existing methods designed natural images encounter sharp performance degradations when they are directly applied to top-view images. Through careful analysis, we observe that challenges mainly come from lack discriminative object features due severe scale variations, low contrasts, and...
Traditional in the wild image quality assessment (IQA) models are generally trained with labels of mean opinion score (MOS), while missing rich subjective information contained ratings, for example, standard deviation scores (SOS) or even distribution (DOS). In this paper, we propose a novel IQA method named RichIQA to explore rating beyond MOS predict wild. is characterized by two key designs: (1) three-stage prediction network which exploits powerful feature representation capability...
We first develop a rate-distortion (R-D) model for DCT-based video coding incorporating the macroblock (MB) intra refreshing rate. For any given bit rate and rate, this is capable of estimating corresponding distortion even before frame coded. then present theoretical analysis picture caused by channel errors subsequent inter-frame propagation. Based on analysis, we statistical to estimate such induced different conditions encoder settings. The proposed analytic mathematically describes...
We proposed compressive data gathering (CDG) that leverages sampling (CS) principle to efficiently reduce communication cost and prolong network lifetime for large scale monitoring sensor networks. The capacity has been proven increase proportionally the sparsity of readings. In this paper, we further address two key problems in CDG framework. First, investigate how generate RIP (restricted isometry property) preserving measurements readings by taking multi-hop into account. Excitingly,...
In this paper, we address a haze removal problem from single nighttime image, even in the presence of varicolored and non-uniform illumination. The core idea lies novel maximum reflectance prior. We first introduce hazy imaging model, which includes local ambient illumination item both direct attenuation term scattering term. Then, propose simple but effective image prior, to estimate varying prior is based on key observation: for most daytime haze-free patches, each color channel has very...
Abstract Background Recently, many studies utilizing next generation sequencing have investigated plant evolution and domestication in annual crops. Peach, Prunus persica , is a typical perennial fruit crop that has ornamental edible varieties. Unlike other crops, cultivated peach includes large number of phenotypes but few polymorphisms. In this study, we explore the genetic basis influence humans on its evolution. Results We perform large-scale resequencing 10 wild 74 varieties, including...
The increasing popularity of video (i.e., audio-visual) applications or services over both wired and wireless links has prompted recent growing interests in the investigations quality experience (QoE) online transmission. Conventional metrics, such as peak-signal-to-noise-ratio service, only focus on reception from systematic perspective. As a result, they cannot represent true visual an individual user. Instead, QoE introduces user experience-driven strategy which puts special emphasis...
This paper surveys the emerging paradigm of cloud mobile media.We start with two alternative perspectives for media networks: an end-to-end view and a layered view.Summaries existing research in this area are organized according to service framework: i) resource management control infrastructure-as-a-service (IaaS), ii) cloud-based services platform-as-a-service (PaaS), iii) novel systems applications software-as-a-service (SaaS).We further substantiate our proposed design principles using...
Unsupervised image translation, which aims in translating two independent sets of images, is challenging discovering the correct correspondences without paired data. Existing works build upon Generative Adversarial Networks (GANs) such that distribution translated images are indistinguishable from target set. However, set-level constraints cannot learn instance-level (e.g. aligned semantic parts object transfiguration task). This limitation often results false positives geometric or...
With the promising applications in e-Health and entertainment services, wireless body area network (WBAN) has attracted significant interest. One critical challenge for WBAN is to track maintain quality of service (QoS), e.g., delivery probability latency, under dynamic environment dictated by human mobility. Another important issue ensure energy efficiency within such a resource-constrained network. In this paper, new medium access control (MAC) protocol proposed tackle these two...
The analysis of polarized filtered images has been proven useful in image dehazing. However, the current polarization-based dehazing algorithms are based on assumption that polarization is only associated with airlight. This does not hold up well practice since both object radiance and airlight contribute to polarization. In this study, a new hazy imaging model presented, which considers joint effects process. addition, an effective method synthesize optimal polarized-difference (PD)...
As the image data produced by individuals and enterprises is rapidly increasing, Scalar Invariant Feature Transform (SIFT), as a local feature detection algorithm, has been heavily employed in various areas, including object recognition, robotic mapping, etc. In this context, there growing need to outsource such computation with high complexity cloud for its economic computing resources on-demand ubiquitous access. However, how protect private while enabling becomes major concern. To address...
The worldwide flourishing of the Internet Things (IoT) in past decade has enabled numerous new applications through internetworking a wide variety devices and sensors. More recently, visual sensors have seen their considerable booming IoT systems because they are capable providing richer more versatile information. Internetworking large-scale been named Video (IoVT). IoVT its own unique characteristics terms sensing, transmission, storage, analysis, which fundamentally different from...
In this paper, we consider power allocation in multiband cognitive radio (CR) networks, where multiple secondary users (SUs) transmit via a common relay and compete for the of relay. We employ hybrid overlay/underlay spectrum sharing scheme, allowing SU to adapt its way accessing licensed status primary user (PU). If PU is detected be idle at selected channel, works an overlay mode; else, it underlay. addition, auction-based power-allocation scheme proposed solve competition SUs. For working...