A5100687159- Image Retrieval and Classification Techniques
- Advanced Image and Video Retrieval Techniques
- Aesthetic Perception and Analysis
- Video Analysis and Summarization
- Visual Attention and Saliency Detection
- Biomedical Text Mining and Ontologies
- Medical Image Segmentation Techniques
- Semantic Web and Ontologies
- Advanced Data Storage Technologies
- Cloud Computing and Resource Management
- Distributed systems and fault tolerance
- Emotion and Mood Recognition
- Peer-to-Peer Network Technologies
- Caching and Content Delivery
- AI in cancer detection
- Image Processing and 3D Reconstruction
- Remote-Sensing Image Classification
- Advanced Neural Network Applications
- Human Pose and Action Recognition
- Bioinformatics and Genomic Networks
- Domain Adaptation and Few-Shot Learning
- IoT and Edge/Fog Computing
- Image and Signal Denoising Methods
- Video Surveillance and Tracking Methods
- Topic Modeling
Imperial College London
2022-2025
Hammersmith Hospital
2025
Pennsylvania State University
2015-2024
Xi'an University of Technology
2024
University of Toronto
2024
University of Wisconsin–Madison
2020-2024
Loughborough University
2024
Clemson University
2013-2023
Columbia University
1980-2023
Federal Reserve Board of Governors
2017-2023
We present here SIMPLIcity (semantics-sensitive integrated matching for picture libraries), an image retrieval system, which uses semantics classification methods, a wavelet-based approach feature extraction, and region based upon segmentation. An is represented by set of regions, roughly corresponding to objects, are characterized color, texture, shape, location. The system classifies images into semantic categories. Potentially, the categorization enhances permitting semantically-adaptive...
Abstract Motivation: Although controlled biochemical or biological vocabularies, such as Gene Ontology (GO) (http://www.geneontology.org), address the need for consistent descriptions of genes in different data sources, there is still no effective method to determine functional similarities based on gene annotation information from heterogeneous sources. Results: To this critical need, we proposed a novel encode GO term's semantics (biological meanings) into numeric value by aggregating...
Automatic linguistic indexing of pictures is an important but highly challenging problem for researchers in computer vision and content-based image retrieval. In this paper, we introduce a statistical modeling approach to problem. Categorized images are used train dictionary hundreds models each representing concept. Images any given concept regarded as instances stochastic process that characterizes the To measure extent association between textual description concept, likelihood occurrence...
Multiple-instance problems arise from the situations where training class labels are attached to sets of samples (named bags), instead individual within each bag (called instances). Most previous multiple-instance learning (MIL) algorithms developed based on assumption that a is positive if and only at least one its instances positive. Although works well in drug activity prediction problem, it rather restrictive for other applications, especially those computer vision area. We propose...
Designing computer programs to automatically categorize images using low-level features is a challenging research topic in vision. In this paper, we present new learning technique, which extends Multiple-Instance Learning (MIL), and its application the problem of region-based image categorization. Images are viewed as bags, each contains number instances corresponding regions obtained from segmentation. The standard MIL assumes that bag labeled positive if at least one positive; otherwise,...
Developing effective methods for automated annotation of digital pictures continues to challenge computer scientists. The capability annotating by computers can lead breakthroughs in a wide range applications, including Web image search, online picture-sharing communities, and scientific experiments. In this work, the authors developed new optimization estimation techniques address two fundamental problems machine learning. These serve as basis automatic linguistic indexing - real time...
The last decade has witnessed great interest in research on content-based image retrieval. This paved the way for a large number of new techniques and systems, growing associated fields to support such systems. Likewise, digital imagery expanded its horizon many directions, resulting an explosion volume data required be organized. In this paper, we discuss some key contributions current related retrieval automated annotation, spanning 120 references. We also challenges involved adaptation...
This paper proposes a fuzzy logic approach, UFM (unified feature matching), for region-based image retrieval. In our retrieval system, an is represented by set of segmented regions, each which characterized (fuzzy set) reflecting color, texture, and shape properties. As result, associated with family features corresponding to regions. Fuzzy naturally characterize the gradual transition between regions (blurry boundaries) within incorporate segmentation-related uncertainties into algorithm....
In this tutorial, we define and discuss key aspects of the problem computational inference aesthetics emotion from images. We begin with a background discussion on philosophy, photography, paintings, visual arts, psychology. This is followed by introduction set problems that research community has been striving to solve framework required for solving them. also describe data sets available performing assessment outline several real-world applications where in domain can be employed. A...
Effective visual features are essential for computational aesthetic quality rating systems. Existing methods used machine learning and statistical modeling techniques on handcrafted or generic image descriptors. A recently-published large-scale dataset, the AVA has further empowered based approaches. We present RAPID (RAting PIctorial aesthetics using Deep learning) system, which adopts a novel deep neural network approach to enable automatic feature learning. The central idea is incorporate...
This paper investigates problems of image style, aesthetics, and quality estimation, which require fine-grained details from high-resolution images, utilizing deep neural network training approach. Existing convolutional networks mostly extracted one patch such as a down-sized crop each example. However, may not always well represent the entire image, cause ambiguity during training. We propose multi-patch aggregation approach, allows us to train models using multiple patches generated...
Metagenomic next-generation sequencing (mNGS) has enabled the rapid, unbiased detection and identification of microbes without pathogen-specific reagents, culturing, or a priori knowledge microbial landscape. mNGS data analysis requires series computationally intensive processing steps to accurately determine composition sample. Existing tools typically require bioinformatics expertise access local server-class hardware resources. For many research laboratories, this presents an obstacle,...
A survey of the literature reveals that image processing tools aimed at supplementing art historian's toolbox are currently in earliest stages development. To jump-start development such methods, Van Gogh and Kroller-Muller museums The Netherlands agreed to make a data set 101 high-resolution gray-scale scans paintings within their collections available groups researchers from several different universities. This article describes approaches brushwork analysis artist identification developed...
Unlike conventional facial expressions, microexpressions are instantaneous and involuntary reflections of human emotion. Because fleeting, lasting only a few frames within video sequence, they difficult to perceive interpret correctly, highly challenging identify categorize automatically. Existing recognition methods often ineffective at handling subtle face displacements, which can be prevalent in typical microexpression applications due the constant movements individuals being observed. To...
This paper investigates unified feature learning and classifier training approaches for image aesthetics assessment . Existing methods built upon handcrafted or generic features developed machine statistical modeling techniques utilizing examples. We adopt a novel deep neural network approach to allow estimate aesthetics. In particular, we develop double-column convolutional support heterogeneous inputs, i.e., global local views, in order capture both characteristics of images addition,...
Model pruning has become a useful technique that improves the computational efficiency of deep learning, making it possible to deploy solutions in resource-limited scenarios. A widely-used practice relevant work assumes smaller-norm parameter or feature plays less informative role at inference time. In this paper, we propose channel for accelerating computations convolutional neural networks (CNNs) does not critically rely on assumption. Instead, focuses direct simplification...
Content-based image retrieval using region segmentation has been an active research area. We present IRM (Integrated Region Matching), a novel similarity measure for region-based comparison. The targeted systems represent by set of regions, roughly corresponding to objects, which are characterized features reflecting color, texture, shape, and location properties. evaluating overall between images incorporates properties all the regions in region-matching scheme. Compared with based on...
In a typical content-based image retrieval (CBIR) system, target images (images in the database) are sorted by feature similarities with respect to query. Similarities among usually ignored. This paper introduces new technique, cluster-based of unsupervised learning (CLUE), for improving user interaction systems fully exploiting similarity information. CLUE retrieves clusters applying graph-theoretic clustering algorithm collection vicinity Clustering is dynamic. particular, formed depend on...
During the four years of AgRISTARS Program, significant progress was made in quantifying capabilities microwave sensors for remote sensing soil moisture. In this paper we discuss results numerous field and aircraft experiments, analysis spacecraft data, modeling activities which examined various noise factors such as roughness vegetation that affect interpretability emission measurements. While determining a 21-cm wavelength radiometer best single sensor moisture research, these studies...
To design a fuzzy rule-based classification system (fuzzy classifier) with good generalization ability in high dimensional feature space has been an active research topic for long time. As powerful machine learning approach pattern recognition problems, the support vector (SVM) is known to have ability. More importantly, SVM can work very well on high- (or even infinite) space. This paper investigates connection between classifiers and kernel machines, establishes link rules kernels,...