A5100626347- Advanced Image and Video Retrieval Techniques
- Image Retrieval and Classification Techniques
- Digital Media Forensic Detection
- Advanced Steganography and Watermarking Techniques
- Video Surveillance and Tracking Methods
- Chaos-based Image/Signal Encryption
- Advanced Data Compression Techniques
- Face and Expression Recognition
- Face recognition and analysis
- Remote-Sensing Image Classification
- Image and Signal Denoising Methods
- Advanced Neural Network Applications
- Video Analysis and Summarization
- Medical Image Segmentation Techniques
- Visual Attention and Saliency Detection
- Multimodal Machine Learning Applications
- Advanced Vision and Imaging
- Robotics and Sensor-Based Localization
- Spectroscopy Techniques in Biomedical and Chemical Research
- Text and Document Classification Technologies
- Human Pose and Action Recognition
- Bioinformatics and Genomic Networks
- Remote Sensing and LiDAR Applications
- Advanced Image Processing Techniques
- Advanced Graph Neural Networks
Utah State University
2014-2024
Yuhuangding Hospital
2015-2023
Qingdao University
2023
Chongqing University of Posts and Telecommunications
2019-2021
Case Western Reserve University
2014
Carleton College
2012
Cleveland Clinic
2008
Beihang University
2008
University of Puget Sound
2008
Cleveland Clinic Lerner College of Medicine
2005
Learning discriminative features for Facial Expression Recognition (FER) in the wild using Convolutional Neural Networks (CNNs) is a non-trivial task due to significant intra-class variations and inter-class similarities. Deep Metric (DML) approaches such as center loss its variants jointly optimized with softmax have been adopted many FER methods enhance power of learned embedding space. However, equally supervising all metric learning method might include irrelevant ultimately degrade...
Bim, the Bcl-2 interacting mediator of cell death, is a member BH3-only family pro-apoptotic proteins. Recent studies have demonstrated that apoptotic activity Bim can be regulated through post-translational mechanism whereby ERK phosphorylation serves as signal for ubiquitination and proteasomal degradation. In this report, we investigated signaling pathways leading to in Ba/F3 cells, an interleukin-3 (IL-3)-dependent B-cell line. IL-3 stimulation induced Bim(EL), one predominant isoforms...
In this work, an algorithm is proposed to scramble JPEG compressed image without causing bitstream size expansion. The causes of expansion in the existing scrambling methods are first identified. Three recommendations on AC coefficients scrambled combat unauthorized viewing. As step algorithm, edges identified directly frequency domain using solely relying any traditional methods. These then form a low resolution its original counterpart and information utilized identify regions. DC encoded...
Breast ultrasound (BUS) imaging is commonly used in the early detection of breast cancer as a portable, valuable, and widely available diagnosis tool. Automated BUS image classification segmentation can assist radiologists making accurate fast decisions. Recent studies illustrate that tumor, peritumoral, background regions images provide valuable information for or classification. However, few have investigated influence these three on multi-task learning. In this study, we propose an...
Facial Expression Recognition (FER) has demonstrated remarkable progress due to the advancement of deep Convolutional Neural Networks (CNNs). FER's goal as a visual recognition problem is learn mapping from facial embedding space set fixed expression categories using supervised learning algorithm. Softmax loss de facto standard in practice fails discriminative features for efficient learning. Center and its variants promising solutions increase feature discriminability enable They...
Ultrasound imaging is one of the most commonly used diagnostic tools to detect and classify abnormalities women breast. Automatic ultrasound image segmentation provides radiologists a second opinion increase diagnosis accuracy. Deep neural networks have recently been employed achieve better results than conventional approaches. In this paper, we propose novel deep learning architecture, Multi-Scale Self-Attention Network (MSSA-Net), which can be trained on small datasets explore...
Abstract Background Existing methods for calculating semantic similarity between gene products using the Gene Ontology (GO) often rely on external resources, which are not part of ontology. Consequently, changes in these resources like biased term distribution caused by shifting hot research topics, will affect calculation similarity. One way to avoid this problem is use that "intrinsic" ontology, i.e. independent knowledge. Results We present a shortest-path graph kernel (spgk) method...
The nanostructures and hydrophobic properties of cancer cell membranes are important for membrane fusion adhesion. They directly related to biophysical properties, including aggressive growth migration. Additionally, chemical component analysis the could potentially be applied in clinical diagnosis by identification specific biomarker receptors expressed on surfaces. In present work, a combined Raman microspectroscopy (RM) atomic force microscopy (AFM) technique was detect difference...
Face recognition under illumination variations is a challenging research area. This paper presents new method based on the log function and fractal analysis (FA) to produce logarithmic dimension (LFD) image which invariant. The proposed FA feature-based very effective edge enhancer technique extract enhance facial features such as eyes, eyebrows, nose, mouth. Our extensive experiments show achieves best accuracy using one per subject for training when compared six recently state-of-the-art methods.
Raman spectroscopy has been playing an increasingly significant role for cell classification. Here, we introduce a novel microfluidic chip non-invasive natural fingerprint collection. Traditional measurement of the cells grown in Polydimethylsiloxane (PDMS) based device suffers from background noise substrate materials PDMS when intended to apply as vitro assay. To overcome this disadvantage, current is designed with middle layer sandwiched by two MgF2 slides which minimize signal...
This study proposes an illumination‐invariant face‐recognition method called adaptive homomorphic eight local directional pattern (AH‐ELDP). AH‐ELDP first uses filtering to reduce the influence of illumination from input face image. It then applies interpolative enhancement function stretch filtered Finally, it produces edge images using Kirsch compass masks and all information create illumination‐insensitive representation. The author's extensive experiments show that technique achieves...
Breast cancer is a great threat to women’s health. Automatic analysis of UltraSound (BUS) images can help radiologists make more accurate and efficient diagnoses breast cancer. We propose Multi-Task Learning Network with Context-Oriented Self-Attention (MTL-COSA) module automatically simultaneously segment tumors classify them as benign or malignant. The COSA incorporates prior medical knowledge guide the network learn contextual relationships for better feature representations in BUS...