A5018695663- Human Pose and Action Recognition
- Gait Recognition and Analysis
- Hand Gesture Recognition Systems
- Domain Adaptation and Few-Shot Learning
- Soft Robotics and Applications
- AI in cancer detection
- Human Motion and Animation
- Multimodal Machine Learning Applications
- Diabetic Foot Ulcer Assessment and Management
- Advanced Neural Network Applications
- Muscle activation and electromyography studies
- Advanced Steganography and Watermarking Techniques
- Stroke Rehabilitation and Recovery
- Prosthetics and Rehabilitation Robotics
- Robotics and Sensor-Based Localization
- Surgical Simulation and Training
- Gaze Tracking and Assistive Technology
- Industrial Vision Systems and Defect Detection
- Advanced Vision and Imaging
- Teleoperation and Haptic Systems
- Retinal Imaging and Analysis
- Anatomy and Medical Technology
- Advanced Image and Video Retrieval Techniques
- Video Surveillance and Tracking Methods
- Radiomics and Machine Learning in Medical Imaging
Shanghai Jiao Tong University
2007-2024
Guizhou University
2024
ShenZhen People’s Hospital
2024
Beijing University of Posts and Telecommunications
2019-2024
Chongqing University of Technology
2023
Southern University of Science and Technology
2023
Defence Electronics Research Laboratory
2022
Imperial College London
2017-2021
City University of Hong Kong
2015-2018
Xidian University
2010-2018
Summary This paper presents a novel model reduction method: deep learning reduced order model, which is based on proper orthogonal decomposition and methods. The approach recent technological advancement in the field of artificial neural networks. It has advantage nonlinear system with multiple levels representation predicting data. In this work, training data are obtained from high fidelity solutions at selected time levels. long short‐term memory network used to construct set hypersurfaces...
Abstract Background Breast cancer causes hundreds of thousands deaths each year worldwide. The early stage diagnosis and treatment can significantly reduce the mortality rate. However, traditional manual needs intense workload, diagnostic errors are prone to happen with prolonged work pathologists. Automatic histopathology image recognition plays a key role in speeding up improving quality diagnosis. Methods In this work, we propose breast classification by assembling multiple compact...
Finite/fixed-time control yields a promising tool to optimize system's settling time, but lacks the ability separately define time and convergence domain (known as <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">practically prescribed-time stability</i> , PPTS). We provide sufficient condition for PPTS based on new piecewise exponential function, which decouples into user-defined parameters. propose an adaptive event-triggered scheme...
Assistive robots play an important role in improving the quality of life patients at home. Among all monitoring tasks, gait disorders are prevalent elderly and people with neurological conditions this increases risk fall. Therefore, development mobile systems for home normal living is important. Here, we present a system that able to track humans analyze their canonical coordinates based on single RGB-D camera. First, view-invariant three-dimensional (3-D) lower limb pose estimation achieved...
For abnormal gait recognition, pattern-specific features indicating abnormalities are interleaved with the subject-specific differences representing biometric traits. Deep representations are, therefore, prone to overfitting, and models derived cannot generalize well new subjects. Furthermore, there is limited availability of data obtained from precise Motion Capture (Mocap) systems because regulatory issues slow adaptation technologies in health care. On other hand, captured markerless...
The recent breakthroughs in computer vision have benefited from the availability of large representative datasets (e.g. ImageNet and COCO) for training. Yet, robotic poses unique challenges applying visual algorithms developed these standard due to their implicit assumption over non-varying distributions a fixed set tasks. Fully retraining models each time new task becomes available is infeasible computational, storage sometimes privacy issues, while naïve incremental strategies been shown...
Monitoring the mental workload of operators is paramount importance in space telerobotic training and other teleoperation tasks. Instead estimation task-specific workload, this article aims at investigating impact two significant confounding factors (time-pressure latency) on explored use eye-tracking technology for factor-induced performance evaluation. Ten subjects teleoperated a Canadarm2 robot to complete complex on-orbit assembly task our photo-realistic simulator while wearing...
For gait analysis, especially for the detection of subtle abnormalities, collected datasets involve high variability across subjects due to inherent biometric traits and movement behaviors, leading limited accuracy poor generalizability. To address this, we propose a novel deep multi-source Unsupervised Domain Adaptation (UDA) approach, namely Maximum Cross-Domain Classifier Discrepancy (MCDCD), which aims improve classification performance on test subject (target domain) by leveraging...
Motion behaviors of a rigid body can be characterized by 6-dimensional motion trajectory, which contains position vectors reference point on the and rotations this over time. This paper devises Rotation Relative Velocity (RRV) descriptor exploring local translational rotational invariants trajectories bodies, is insensitive to noise, invariant transformation scaling. A flexible metric also introduced measure distance between two RRV descriptors. The then applied characterize motions human...
Accurate real-time catheter segmentation is an important pre-requisite for robot-assisted endovascular intervention. Most of the existing learning-based methods and tracking are only trained on smallscale datasets or synthetic data due to difficulties ground-truth annotation. Furthermore, temporal continuity in intraoperative imaging sequences not fully utilised. In this paper, we present FW-Net, end-to-end deep learning framework The proposed FW-Net has three modules: a network with...
Holographic acoustic field has shown great potential for non-contact robotic manipulations of millimeter or sub-millimeter size objects to effectively deliver power. The latest technology generating dynamic holographic is through phased transducer array, where relative phases emitted waves from transducers are independently controlled modulate the interference field. While forward kinematics a array based manipulation system simple and straightforward, inverse (i.e., mapping given control...
Egocentric vision has gained increasing popularity recently, opening new avenues for human-centric applications. However, the use of egocentric fisheye cameras allows wide angle coverage but image distortion is introduced along with strong human body self-occlusion imposing significant challenges in data processing and model reconstruction. Unlike previous work only leveraging synthetic training, this paper presents a real-world EgoCentric Human Pose (ECHP) dataset. To tackle difficulty...
Photoacoustic microscopy (PAM) has gained increasing popularity in biomedical imaging, providing new opportunities for tissue monitoring and characterization. With the development of deep learning techniques, convolutional neural networks have been used PAM image resolution enhancement denoising. However, there exist several inherent challenges this approach. This work presents a
Advances in depth sensing technologies have allowed simultaneous acquisition of both color and data under different environments. However, most sensors lower resolution than that the associated channels such a mismatch can affect applications require accurate recovery. Existing enhancement methods use simplistic noise models cannot generalize well real-world conditions. In this paper, coupled real-synthetic domain adaptation method is proposed, which enables transfer between high-quality...
A novel video steganography scheme based on motion vectors and linear block codes has been proposed in this paper. Our method embed secret messages the of cover media during process H.264 compressing. Linear used to reducing modification rate vectors. The steganographic is not only lowly computational complexity, but also highly imperceptible human being. Furthermore, information can be extracted directly without using original sequences. Experiments are designed prove feasibility method....
The early detection of gait abnormalities plays a key role in medical applications, where most the previous abnormal recognition methods rely on kinematic data captured with vision-based systems or wearable inertial sensors. This paper, conversely, puts forward ambitious objective to employ multiple Electromyography (EMG) sensors for detection. Our proposed approach uses eight wireless EMG attached skin electrodes four muscles (i.e., Tibialis Anterior, Peroneus Longus, Gas-trocnemius, and...
Attention-deficit/hyperactivity disorder (ADHD) is a common neurodevelopmental in school-aged children. The lack of objective biomarkers for ADHD often results missed diagnoses or misdiagnoses, which lead to inappropriate delayed interventions. Eye-tracking technology provides an method assess children's neuropsychological behavior.
Accurate master-slave control is important for Robot-Assisted Microsurgery (RAMS). This paper presents a handheld master controller the operation and training of RAMS. A 9-axis Inertial Measure Unit (IMU) micro camera are utilized to form sensing system controller. new hybrid marker pattern designed achieve reliable visual tracking, which integrated QR codes, Aruco markers, chessboard vertices. Real-time multi-sensor fusion implemented further improve tracking accuracy. The proposed has been...
As increasing attention is paid to human action recognition from skeleton data, this paper focuses on such tasks by proposing a hierarchical model discover the structure information of body-parts involved in actions for better analysis data. Considering as simultaneous motions skeleton, we propose simultaneously apply discriminative selection at same scale and group coupling bundles different scales, while decompose into hierarchy varying scales. To represent body-parts, accordingly build...