A5013527575- Video Analysis and Summarization
- Advanced Image and Video Retrieval Techniques
- Image Retrieval and Classification Techniques
- Music and Audio Processing
- Image Processing Techniques and Applications
- Human Pose and Action Recognition
- Anomaly Detection Techniques and Applications
- Multimodal Machine Learning Applications
- Cell Image Analysis Techniques
- Context-Aware Activity Recognition Systems
- Advanced Vision and Imaging
- Video Surveillance and Tracking Methods
- Genomics and Phylogenetic Studies
- Natural Language Processing Techniques
- EEG and Brain-Computer Interfaces
- Data Mining Algorithms and Applications
- Data Management and Algorithms
- Time Series Analysis and Forecasting
- Gaze Tracking and Assistive Technology
- Visual Attention and Saliency Detection
- Medical Image Segmentation Techniques
- AI in cancer detection
- Speech and Audio Processing
- Data Visualization and Analytics
- Rough Sets and Fuzzy Logic
Kindai University
2020-2023
Doshisha University
2023
University of Siegen
2013-2018
Kobe University
2005-2013
Muroran Institute of Technology
2012
Getting a good feature representation of data is paramount for Human Activity Recognition (HAR) using wearable sensors. An increasing number learning approaches-in particular deep-learning based-have been proposed to extract an effective by analyzing large amounts data. However, getting objective interpretation their performances faces two problems: the lack baseline evaluation setup, which makes strict comparison between them impossible, and insufficiency implementation details, can hinder...
The scarcity of labelled time-series data can hinder a proper training deep learning models. This is especially relevant for the growing field ubiquitous computing, where coming from wearable devices have to be analysed using pattern recognition techniques provide meaningful applications. To address this problem, we propose transfer method based on attributing sensor modality labels large amount collected various application fields. Using these data, our firstly trains Deep Neural Network...
The analysis of sleep stages for children plays an important role in early diagnosis and treatment. This paper introduces our stage classification method addressing the following two challenges: first is data imbalance problem, i.e., highly skewed class distribution with underrepresented minority classes. For this, a Gaussian Noise Data Augmentation (GNDA) algorithm was applied to polysomnography recordings seek balance sizes different stages. second challenge difficulty identifying stages,...
The Gamma-Ray and AntiMatter Survey (GRAMS) is a next-generation balloon/satellite mission utilizing Liquid Argon Time Projection Chamber (LArTPC) detector to measure both MeV gamma rays antinuclei produced by dark matter annihilation or decay. GRAMS can identify antihelium-3 events based on the measurements of X-rays charged pions from decay exotic atoms, Flight (TOF), energy deposition, stopping range. This paper shows sensitivity estimation using GEANT4 Monte Carlo simulation. For...
Artificial intelligence and especially deep learning methods have achieved outstanding results for various applications in the past few years. Pain recognition is one of them, as models been proposed to replace previous gold standard with an automated objective assessment. While accuracy such could be increased incrementally, understandability transparency these systems not main focus research community thus far. Thus, this work, several outcomes insights explainable artificial applied...
This paper addresses wearable-based recognition of Activities Daily Living (ADLs) which are composed several repetitive and concurrent short movements having temporal dependencies. It is improbable to directly use sensor data recognize these long-term composite activities because two examples (data sequences) the same ADL result in largely diverse sensory data. However, they may be similar terms more semantic meaningful short-term atomic actions. Therefore, we propose a two-level...
Most human detection algorithms in depth images perform well detecting and tracking the movements of a single object. However, their performance is rather poor when person occluded by other objects or there are multiple humans present scene. In this paper, we propose novel technique which analyzes edges image to detect people. The proposed detects head through fast template matching algorithm verifies it 3D model fitting technique. entire body extracted from using simple segmentation scheme...
With the recent spread of mobile devices equipped with different sensors, it is possible to continuously recognise and monitor activities in daily life. This sensor-based human activity recognition formulated as sequence classification categorise sequences sensor values into appropriate classes. One crucial problem how model features that can precisely represent characteristics each lead accurate recognition. It laborious and/or difficult hand-craft such based on prior knowledge manual...
Environmental Microorganisms (EMs) are currently recognised using molecular biology (DNA, RNA) or morphological methods. The first ones very time-consuming and expensive. second require a experienced laboratory operator. To overcome these problems, we introduce an automatic classification method for EMs in the framework of content-based image analysis this paper. describe shapes observed microscopic images, use Edge Histograms, Fourier Descriptors, extended Geometrical Features, as well...
Occurrence of certain environmental microorganisms and their species is a very informative indicator to evaluate quality. Unfortunately, manual recognition in microbiological laboratories time-consuming expensive. Therefore, we work on an automatic method for shape-based classification EMs microscopic images. First, segment the from background. Second, describe shapes by discriminative feature vectors. Third, perform EM using Support Vector Machines. The most important scientific...
This paper develops a query-by-example method for retrieving shots of an event (event shots) using example provided by user. The following three problems are mainly addressed. Firstly, cannot be retrieved single model as they contain significantly different features due to varied camera techniques, settings and so forth. is overcome rough set theory extract multiple classification rules with each rule specialized retrieve portion shots. Secondly, since user can only provide small number...
Environmental Microorganisms (EMs), such as Epistylis and Rotifera, are very tiny living beings in human environments decompose pollutants their nutrition. The classification of EMs plays a fundamental role for establishing sustainable ecosystem. However, this is traditionally done by molecular methods using DNA or RNA analysis, morphological microscopes. These require huge monetary costs manual efforts. In paper, we propose an EM method directly processing microscopic images. We especially...
One crucial problem in sensor-based human activity recognition is how to model features that can precisely represent characteristics of a sequence sensor values. For this, we study codebook approach represents the as distribution characteristic subsequences. The extensive experiments on different tasks for physical, mental and eye-based activities validate effectiveness, generality usability approach, where only few intuitive parameters need be tuned.