A5080054465- Cell Image Analysis Techniques
- Video Surveillance and Tracking Methods
- Human Pose and Action Recognition
- Image Processing Techniques and Applications
- Advanced Neural Network Applications
- Digital Holography and Microscopy
- Hand Gesture Recognition Systems
- AI in cancer detection
- COVID-19 diagnosis using AI
- Advanced Image and Video Retrieval Techniques
- Anomaly Detection Techniques and Applications
- Domain Adaptation and Few-Shot Learning
- Advanced Vision and Imaging
- Multimodal Machine Learning Applications
- Infrastructure Maintenance and Monitoring
- 3D Surveying and Cultural Heritage
- Context-Aware Activity Recognition Systems
- Autonomous Vehicle Technology and Safety
- Indoor and Outdoor Localization Technologies
- Consumer Market Behavior and Pricing
- Industrial Vision Systems and Defect Detection
- Traffic Prediction and Management Techniques
- Advanced X-ray Imaging Techniques
- Visual Attention and Saliency Detection
- Digital Transformation in Industry
Stony Brook University
2020-2024
State University of New York
2021
Missouri University of Science and Technology
2012-2020
China Southern Power Grid (China)
2020
Alibaba Group (United States)
2019
Bellevue Hospital Center
2019
Carnegie Mellon University
2010-2018
University of Nottingham
2011
Pennsylvania State University
2007-2009
University of Wisconsin–Madison
2004
Human physical activity recognition based on wearable sensors has applications relevant to our daily life such as healthcare. How achieve high accuracy with low computational cost is an important issue in the ubiquitous computing. Rather than exploring handcrafted features from time-series sensor signals, we assemble signal sequences of accelerometers and gyroscopes into a novel image, which enables Deep Convolutional Neural Networks (DCNN) automatically learn optimal image for task. Our...
American Sign Language (ASL) alphabet recognition using marker-less vision sensors is a challenging task due to the complexity of ASL signs, self-occlusion hand, and limited resolution sensors. This paper describes new method for low-cost depth camera, which Microsoft's Kinect. A segmented hand configuration first obtained by contrast feature based per-pixel classification algorithm. Then, hierarchical mode-seeking developed implemented localize joint positions under kinematic constraints....
Although recent works in semi-supervised learning (SemiSL) have accomplished significant success natural image segmentation, the task of discriminative representations from limited annotations has been an open problem medical images. Contrastive Learning (CL) frameworks use notion similarity measure which is useful for classification problems, however, they fail to transfer these quality accurate pixel-level segmentation. To this end, we propose a novel patch-based CL framework segmentation...
In a smart manufacturing system involving workers, recognition of the worker’s activity can be used for quantification and evaluation performance, as well to provide onsite instructions with augmented reality. this paper, we propose method using Inertial Measurement Unit (IMU) surface electromyography (sEMG) signals obtained from Myo armband. The raw 10-channel IMU are stacked form signal image. This image is transformed into an by applying Discrete Fourier Transformation (DFT) then fed...
The rapid advancement of sensor technologies and artificial intelligence are creating new opportunities for traffic safety enhancement. Dashboard cameras (dashcams) have been widely deployed on both human driving vehicles automated vehicles. A computational model that can accurately promptly predict accidents from the dashcam video will enhance preparedness accident prevention. spatial-temporal interaction agents is complex. Visual cues predicting a future embedded deeply in data. Therefore,...
We present several algorithms for cell image analysis including microscopy restoration, event detection and tracking in a large population. The are integrated into an automated system capable of quantifying proliferation metrics vitro real-time. This offers unique opportunities biological applications such as efficient behavior discovery response to different culturing conditions adaptive experiment control. quantitatively evaluated our system's performance on 16 sequences with satisfactory...
Automated cell tracking in populations is important for research and discovery biology medicine. In this paper, we propose a method based on global spatio-temporal data association which considers hypotheses of initialization, termination, translation, division false positive an integrated formulation. Firstly, reliable tracklets (i.e., short trajectories) are generated by linking detection responses frame-by-frame association. Next, these globally associated over time to obtain final...
Production innovations are occurring faster than ever. Manufacturing workers thus need to frequently learn new methods and skills. In fast changing, largely uncertain production systems, manufacturers with the ability comprehend workers’ behavior assess their operation performance in near real-time will achieve better peers. Action recognition can serve this purpose. Despite that human action has been an active field of study machine learning, limited work done for recognizing worker actions...
We propose weakly supervised training schemes to train end-to-end cell segmentation networks that only require a single point annotation per as the label and generate high-quality mask close those fully methods using on cells. Three are investigated networks, annotation. First, self-training is performed learn additional information near annotated points. Next, co-training applied more regions multiple supervise each other. Finally, hybrid-training scheme proposed leverage advantages of both...
Abstract The quantification of repetitive movements, known as action counting, is critical in various applications, such fitness tracking, rehabilitation, and manufacturing operation monitoring. Traditional methods predominantly relied on the estimation red-green-and-blue (RGB) frames body pose landmarks to identify number repetitions. However, these suffer from several issues, instability under varying camera viewpoints, propensity for over-counting or under-counting, challenges...
Cell segmentation in microscopy imagery is essential for many bioimage applications such as cell tracking. To segment cells from the background accurately, we present a pixel classification approach that independent of type or imaging modality. We train set Bayesian classifiers clustered local training image patches. Each classifier an expert to make decision its specific domain. The mixture experts determines how likely new pixel. demonstrate effectiveness this on four types with diverse...
Automatic cell segmentation can hardly be flawless due to the complexity of image data particularly when time-lapse experiments last for a long time without biomarkers. To address this issue, we propose an interactive method by classifying feature-homogeneous superpixels into specific classes, which is guided human interventions. Specifically, actively select most informative minimizing expected prediction error upper bounded transductive Rademacher complexity, and then query annotations....
In 2019, outbreaks of vaccine-preventable diseases reached the highest number in US since 1992. Medical misinformation, such as antivaccine content propagating through social media, is associated with increases vaccine delay and refusal. Our overall goal to develop an automatic detector for messages counteract negative impact that have on public health. Very few extant detection systems considered multimodality media posts (images, texts, hashtags), instead focus textual components, despite...
Abstract Phase contrast time-lapse microscopy is a non-destructive technique that generates large volumes of image-based information to quantify the behaviour individual cells or cell populations. To guide development algorithms for computer-aided tracking and analysis, 48 image sequences, each spanning approximately 3.5 days, were generated with accompanying ground truths C2C12 myoblast cultured under 4 different media conditions, including fibroblast growth factor 2 (FGF2), bone...
Abstract As artificial intelligence and industrial automation are developing, human–robot collaboration (HRC) with advanced interaction capabilities has become an increasingly significant area of research. In this paper, we design develop a real-time, multi-model HRC system using speech gestures. A set 16 dynamic gestures is designed for communication from human to robot. data constructed, it will be shared the community. convolutional neural network developed recognize in real time motion...