A5031773367- Face recognition and analysis
- Face and Expression Recognition
- Medical Image Segmentation Techniques
- Biometric Identification and Security
- Human Pose and Action Recognition
- Video Surveillance and Tracking Methods
- Cardiac Imaging and Diagnostics
- Advanced Image and Video Retrieval Techniques
- Anomaly Detection Techniques and Applications
- Image and Signal Denoising Methods
- Advanced Vision and Imaging
- Cell Image Analysis Techniques
- Coronary Interventions and Diagnostics
- 3D Shape Modeling and Analysis
- Cardiovascular Disease and Adiposity
- Image Retrieval and Classification Techniques
- Robotics and Sensor-Based Localization
- Advanced Fluorescence Microscopy Techniques
- Photoacoustic and Ultrasonic Imaging
- Image Processing Techniques and Applications
- AI in cancer detection
- Advanced X-ray and CT Imaging
- Domain Adaptation and Few-Shot Learning
- Gait Recognition and Analysis
- Cardiovascular Function and Risk Factors
University of Houston
2016-2025
Cornell University
2008-2024
The University of Texas Health Science Center at Houston
2024
Stanford Medicine
2024
Stanford University
2024
Icahn School of Medicine at Mount Sinai
2024
National Heart Lung and Blood Institute
2024
University of Louisville Hospital
2024
University of California, Irvine
2024
Australian National University
2024
In this paper, we present the computational tools and a hardware prototype for 3D face recognition. Full automation is provided through use of advanced multistage alignment algorithms, resilience to facial expressions by employing deformable model framework, invariance capture devices suitable preprocessing steps. addition, scalability in both time space achieved converting scans into compact metadata. We our results on largest known, now publicly available, Face Recognition Grand Challenge...
Monocular 3D facial shape reconstruction from a single 2D image has been an active research area due to its wide applications. Inspired by the success of deep neural networks (DNN), we propose DNN-based approach for End-to-End FAce Reconstruction (UH-E2FAR) image. Different recent works that reconstruct and refine face in iterative manner using both RGB initial rendering, our DNN model is end-to-end, thus complicated rendering process can be avoided. Moreover, integrate architecture two...
For many computer vision applications such as image captioning, visual question answering, and person search, learning discriminative feature representations at both text level is an essential yet challenging problem. Its challenges originate from the large word variance in domain well difficulty of accurately measuring distance between features two modalities. Most prior work focuses on latter challenge, by introducing loss functions that help network learn better but fail to account for...
Background Studies have demonstrated that the current US guidelines based on American College of Cardiology/American Heart Association (ACC/AHA) Pooled Cohort Equations Risk Calculator may underestimate risk atherosclerotic cardiovascular disease ( CVD ) in certain high-risk individuals, therefore missing opportunities for intensive therapy and preventing events. Similarly, overestimate low populations resulting unnecessary statin therapy. We used Machine Learning ML to tackle this problem....
We present a new method for the 3D model-based tracking of human body parts. To mitigate difficulties arising due to occlusion among parts, we employ multiple calibrated cameras in mutually orthogonal configuration. In addition, develop criteria time varying active selection set track motion particular part. particular, at every frame, each camera tracks number parts depending on visibility these and observability their predicted from specific camera. relate points occluding contours models...
The uncontrolled conditions of real-world biometric applications pose a great challenge to any face recognition approach. unconstrained acquisition data from uncooperative subjects may result in facial scans with significant variations along the yaw axis. Such can cause extensive occlusions, resulting missing data. In this paper, novel 3D method is proposed that uses symmetry handle variations. It employs an automatic landmark detector estimates and detects occluded areas for each scan....
A 3D landmark detection method for facial scans is presented and thoroughly evaluated. The main contribution of the automatic pose-invariant landmarks on under large yaw variations (that often result in missing data), its robustness against expressions. Three-dimensional information exploited by using local shape descriptors to extract candidate points. include index, a continuous map principal curvature values object's surface, spin images, point distribution. are identified labeled...
We present a novel motion-based approach for the part determination and shape estimation of human's body parts. The novelty technique is that neither prior model human employed nor segmentation assumed. identification strategy (HBPIS) recovers all parts moving based on spatiotemporal analysis its deforming silhouette. formalize process simultaneous 2D by employing supervisory control theory discrete event systems. In addition, in order to acquire 3D parts, we new algorithm which selectively...
We present an automated left ventricular (LV) myocardial boundary extraction method. Automatic localization of the LV is achieved using a motion map and expectation maximization algorithm. The region then segmented intensity-based fuzzy affinity contours are extracted by cost minimization through dynamic programming approach. results from algorithm compared against experienced radiologists Bland Altman analysis were found to have consistent mean bias 7% limits agreement comparable...
This survey focuses on discrete expression classification and facial action unit recognition performed using 3D face data, possibly including a corresponding 2D texture image. Research trends to date are summarized the limitations of current methods discussed. The challenges towards development more accurate automated identified. We also call for standardized experimental protocols in order draw fair meaningful comparisons between different systems.
In this paper, we first offer an overview of advances in the field distance metric learning. Then, empirically compare selected methods using a common experimental protocol. The number learning algorithms proposed keeps growing due to their effectiveness and wide application. However, existing surveys are either outdated or they focus only on few methods. As result, there is increasing need summarize obtained knowledge concise, yet informative manner. Moreover, do not conduct comprehensive...
Diabetes is a life-altering medical condition that affects millions of people and results in many hospitalizations per year. Consequently, predicting the length stay in-hospital diabetic patients has become increasingly important for staffing resource planning. Although statistical methods have been used to predict hospitalized patients, powerful machine learning techniques not yet explored. In this paper, we compare discuss performance various supervised algorithms (i.e., Multiple linear...
In this paper, we explore global and local features obtained from Convolutional Neural Networks (CNN) for learning to estimate head pose localize landmarks jointly. Because there is a high correlation between landmark locations, the distributions reference database learned deep patch are used reduce error in estimation face alignment tasks. First, train GNet on detected region obtain rough of seven primary landmarks. The most similar shape selected initialization pool constructed training...