A5021505567- Nutritional Studies and Diet
- Advanced Chemical Sensor Technologies
- Advanced Vision and Imaging
- Human Pose and Action Recognition
- Video Surveillance and Tracking Methods
- Image Processing Techniques and Applications
- Phonocardiography and Auscultation Techniques
- Image Enhancement Techniques
- Non-Invasive Vital Sign Monitoring
- Machine Learning in Healthcare
- Context-Aware Activity Recognition Systems
- Diet and metabolism studies
- Artificial Intelligence in Healthcare and Education
- Advanced Neural Network Applications
- Olfactory and Sensory Function Studies
- EEG and Brain-Computer Interfaces
- 3D Shape Modeling and Analysis
- Surgical Simulation and Training
- Soft Robotics and Applications
- Gait Recognition and Analysis
- Autonomous Vehicle Technology and Safety
- Robotics and Sensor-Based Localization
- Blind Source Separation Techniques
- Computer Graphics and Visualization Techniques
- Face recognition and analysis
Imperial College London
2018-2025
Stellenbosch University
2017
University of Pittsburgh
2017
Chinese University of Hong Kong
2016
With the increasing demands on quality healthcare and raising cost of care, pervasive is considered as a technological solutions to address global health issues. In particular, recent advances in Internet Things have led development Medical (IoMT). Although such low sensing devices could potentially transform current reactive care preventative security privacy issues system are often overlooked. As medical capture process very sensitive personal data, their associated communications be...
Medical image analysis plays an important role in clinical diagnosis. In this paper, we examine the recent Segment Anything Model (SAM) on medical images, and report both quantitative qualitative zero-shot segmentation results nine benchmarks, covering various imaging modalities, such as optical coherence tomography (OCT), magnetic resonance (MRI), computed (CT), well different applications including dermatology, ophthalmology, radiology. Those benchmarks are representative commonly used...
Automatic food recognition is the very first step towards passive dietary monitoring. In this paper, we address problem of by mining discriminative regions. Taking inspiration from Adversarial Erasing, a strategy that progressively discovers object regions for weakly supervised semantic segmentation, propose novel network architecture in which primary maintains base accuracy classifying an input image, auxiliary adversarially mines regions, and region classifies resulting mined The global...
Conventional approaches to dietary assessment are primarily grounded in self-reporting methods or structured interviews conducted under the supervision of dietitians. These methods, however, often subjective, potentially inaccurate, and time-intensive. Although artificial intelligence (AI)-based solutions have been devised automate process, prior AI methodologies tackle a fragmented landscape (e.g., merely recognizing food types estimating portion size), encounter challenges their ability...
An objective dietary assessment system can help users to understand their behavior and enable targeted interventions address underlying health problems. To accurately quantify intake, measurement of the portion size or food volume is required. For estimation, previous research studies mostly focused on using model-based stereo-based approaches which rely manual intervention require capture multiple frames from different viewing angles be tedious. In this paper, a view synthesis approach...
Dietary assessment is an important tool for nutritional epidemiology studies. To assess the dietary intake, common approach to carry out 24-h recall (24HR), a structured interview conducted by experienced dietitians. Due unconscious biases in such self-reporting methods, many research works have proposed use of vision-based approaches provide accurate and objective assessments. In this article, novel method based on real-time three-dimensional (3-D) reconstruction deep learning view...
Camera-based passive dietary intake monitoring is able to continuously capture the eating episodes of a subject, recording rich visual information, such as type and volume food being consumed, well behaviours subject. However, there currently no method that incorporate these clues provide comprehensive context from (e.g., subject sharing with others, what eating, how much left in bowl). On other hand, privacy major concern while egocentric wearable cameras are used for capturing. In this...
Reinforcement Learning (RL) based control algorithms can learn the strategies for nonlinear and uncertain environment during interacting with it. Guided by rewards generated environment, a RL agent strategy directly in model-free way instead of investigating dynamic model environment. In paper, we propose sampled-data to reduce computational demand. strategy, whole system is hybrid structure, which plant continuous structure while controller (RL agent) adopts discrete structure. Given that...
<ns3:p>Introduction Current dietary assessment methods struggle to accurately capture individuals’ habits. The ‘Standardised and Objective Dietary Intake Assessment Tool’ (SODIAT)-1 study aims assess the effectiveness of three emerging technologies (urine capillary blood biomarkers, wearable camera technology) two online self-reporting tools monitor intake. Methods This randomised controlled crossover trial was conducted at sites (Hammersmith Hospital University Reading) aimed recruit 30 UK...
A novel food volume measurement technique is proposed in this paper for accurate quantification of the daily dietary intake user. The based on simultaneous localisation and mapping (SLAM), a modified version convex hull algorithm, 3D mesh object reconstruction technique. This explores feasibility applying SLAM techniques continuous with monocular wearable camera. sparse map will be generated by after capturing images item camera multiple algorithm applied to form object. target can then...
A novel vision-based approach for estimating individual dietary intake in food sharing scenarios is proposed this paper, which incorporates detection, face recognition and hand tracking techniques. The method validated using panoramic videos capture subjects' eating episodes. results demonstrate that the able to reliably estimate of each as well sequence. To identify items ingested by subject, a transfer learning designed. 4, 200 images with segmentation masks, among 1,500 are newly...
Accurate depth estimation under adverse night conditions has practical impact and applications, such as on autonomous driving rescue robots. In this work, we studied monocular at time in which various weather, light, different road exist, with data captured both RGB event modalities. Event camera can better capture intensity changes by virtue of its high dynamic range (HDR), is particularly suitable to be applied the amount light limited scene. Although retain visual perception that...
A novel blood pressure estimation method based on long short-term memory neural network, one of the recurrent networks being commonly used nowadays, is proposed in this paper for better chronic diseases monitoring. Along with a newly ambulatory (ABP) processing technique called Two-stage Zero-order Holding (TZH) algorithm has also been presented paper. The methodology advantages over traditional algorithms which are Pulse Transit time (PTT). addresses effectiveness by computing...
Assessing dietary intake in epidemiological studies are predominantly based on self-reports, which subjective, inefficient, and also prone to error. Technological approaches therefore emerging provide objective assessments. Using only egocentric videos, this work aims accurate estimation individual through recognizing consumed food items counting the number of bites taken. This is different from previous that rely inertial sensing count bites, recognize visible but not ones. As a subject may...
Optical microrobots have a wide range of applications in biomedical research for both vitro and vivo studies. In most microrobotic systems, the video captured by monocular camera is only way visualizing movements microrobots, planar motion, general, can be system. Accurate depth estimation essential 3D reconstruction or autofocusing microplatforms, while pose are necessary to enhance perception systems enable dexterous micromanipulation other tasks. this paper, we propose data-driven method...
Dietary assessment system has proven as an effective tool to evaluate the eating behavior of patients suffering from diabetes and obesity. To assess dietary intake, traditional method is carry out a 24-hour recall (24HR), structured interview aimed at capturing information on food items portion size consumed by participants. However, unconscious biases are developed easily due individual's subjective perception in this self-reporting technique which may lead inaccuracy. Thus, paper proposed...
In this paper, we address the problem of forecasting trajectory an egocentric camera wearer (ego-person) in crowded spaces. The ability learned from data different wearers walking around real world can be transferred to assist visually impaired people navigation, as well instill human navigation behaviours mobile robots, enabling better human-robot interactions. To end, a novel dataset was constructed, containing trajectories navigating spaces wearing camera, extracted rich contextual data....
Extracting human attributes, such as gender and age, from biometrics have received much attention in recent years. Gender age recognition can provide crucial information for applications security, healthcare, gaming. In this paper, a novel deep learning approach on using single inertial sensors is proposed. The proposed tested the largest available sensor-based gait database with data collected more than 700 subjects. To demonstrate robustness effectiveness of approach, 10 trials...
Device authentication, encryption, and key distribution are of vital importance to any Internet Things (IoT) systems, such as the new smart city infrastructures. This is due concern that attackers could easily exploit lack strong security in IoT devices gain unauthorized access system or hijack perform denial-of-service attacks on other networks. With rise fog edge computing increasing numbers have been equipped with capabilities data analysis deep learning technologies. Deep can be deployed...
Background: Acute upper gastrointestinal bleeding (AUGIB) is a major cause of morbidity and mortality. This presentation however not universally high risk as only 20–30% bleeds require urgent haemostatic therapy. Nevertheless, the current standard care for all patients admitted to an inpatient bed undergo endoscopy within 24 h stratification which invasive, costly difficult achieve in routine clinical practice. Objectives: To develop novel non-endoscopic machine learning models AUGIB predict...
<ns3:p>Introduction Current dietary assessment methods face challenges in accurately capturing individuals’ habits, undermining the efficacy of public health strategies. The ‘Standardised and Objective Dietary Intake Assessment Tool’ (SODIAT)-1 study aims to assess effectiveness three emerging technologies (urine capillary blood biomarkers, wearable camera technology) two online self-reporting tools monitor intake. Methods This randomised controlled crossover trial will recruit 30...
Surgical performance has been shown to be directly related patient outcomes. There is significant variation in surgical and therefore a need measure operative skill accurately reliably. Despite this, current means of assessment rely on expert observation which labor-intensive, prone rater bias unreliable. We present an automatic approach through the tracking instruments endoscopic video. annotate spatial bounds 2600 images use this new dataset train Mask R-CNN, state-of-the-art instance...