A5088225634- COVID-19 diagnosis using AI
- Sparse and Compressive Sensing Techniques
- Synthetic Aperture Radar (SAR) Applications and Techniques
- Anomaly Detection Techniques and Applications
- AI in cancer detection
- Remote-Sensing Image Classification
- Soil Moisture and Remote Sensing
- Radiomics and Machine Learning in Medical Imaging
- Advanced SAR Imaging Techniques
- Video Surveillance and Tracking Methods
- Remote Sensing in Agriculture
- Speech and Audio Processing
- Indoor and Outdoor Localization Technologies
- Wireless Communication Security Techniques
- Cardiac Imaging and Diagnostics
- Advanced Image Processing Techniques
- Domain Adaptation and Few-Shot Learning
- Advanced Steganography and Watermarking Techniques
- Advanced X-ray and CT Imaging
- Remote Sensing and Land Use
- Fire Detection and Safety Systems
- Image and Signal Denoising Methods
- Machine Learning and ELM
- Computational Drug Discovery Methods
- Digital Imaging for Blood Diseases
Tampere University
2019-2025
Tampere University of Applied Sciences
2021
İzmir University of Economics
2017
Coronavirus disease (COVID-19) has been the main agenda of whole world ever since it came into sight. X-ray imaging is a common and easily accessible tool that great potential for COVID-19 diagnosis prognosis. Deep learning techniques can generally provide state-of-the-art performance in many classification tasks when trained properly over large data sets. However, scarcity be crucial obstacle using them detection. Alternative approaches such as representation-based [collaborative or sparse...
Computer-aided diagnosis has become a necessity for accurate and immediate coronavirus disease 2019 (COVID-19) detection to aid treatment prevent the spread of virus. Numerous studies have proposed use Deep Learning techniques COVID-19 diagnosis. However, they used very limited chest X-ray (CXR) image repositories evaluation with small number, few hundreds, samples. Moreover, these methods can neither localize nor grade severity infection. For this purpose, recent explore activation maps...
Coronavirus disease 2019 (COVID-19) has rapidly become a global health concern after its first known detection in December 2019. As result, accurate and reliable advance warning system for the early diagnosis of COVID-19 now priority. The stages is not straightforward task from chest X-ray images according to expert medical doctors because traces infection are visible only when progressed moderate or severe stage. In this study, our aim evaluate ability recent <italic...
The increasing frequency and intensity of droughts heat waves driven by climate change have led to a significant increase in tree mortality worldwide. However, the lack accurate consistent data on location, timing, species, structure dead trees across vast geographical areas limits our understanding climate-induced mortality. Furthermore, standing dying are crucial indicators forest health biodiversity but often overlooked existing resource mapping systems.To address this, we present novel...
Recent advances in intelligent surveillance systems have enabled a new era of smart monitoring wide range applications from health to homeland security. However, this boom data gathering, analyzing and sharing brings also significant privacy concerns. We propose Compressive Sensing (CS) based encryption that is capable both obfuscating selected sensitive parts documents compressively sampling, hence encrypting non-sensitive the document. The scheme uses hiding technique on CS-encrypted...
Accurate land use/land cover classification of synthetic aperture radar (SAR) images plays an important role in environmental, economic, and nature related research areas applications. When fully polarimetric SAR data is not available, single- or dual-polarization can also be used whilst posing certain difficulties. For instance, traditional Machine Learning (ML) methods generally focus on finding more discriminative features to overcome the lack information due dual-polarimetry. Beside...
Classification of polarimetric synthetic aperture radar (PolSAR) images is an active research area with a major role in environmental applications. The traditional Machine Learning (ML) methods proposed this domain generally focus on utilizing highly discriminative features to improve the classification performance, but task complicated by well-known "curse dimensionality" phenomena. Other approaches based deep Convolutional Neural Networks (CNNs) have certain limitations and drawbacks, such...
Security monitoring via ubiquitous cameras and their more extended in intelligent buildings stand to gain from advances signal processing machine learning. While these innovative ground-breaking applications can be considered as a boon, at the same time they raise significant privacy concerns. In fact, recent GDPR (General Data Protection Regulation) legislation has highlighted become an incentive for privacy-preserving solutions. Typical video schemes address concerns by either anonymizing...
In this study, we propose a novel approach to predict the distances of detected objects in an observed scene. The proposed modifies recently Convolutional Support Estimator Networks (CSENs). CSENs are designed compute direct mapping for Estimation (SE) task representation-based classification problem. We further and demonstrate that methods (sparse or collaborative representation) can be used well-designed regression problems especially over scarce data. To best our knowledge, is first...
Hyperspectral image (HSI) classification is an important task in many applications, such as environmental monitoring, medical imaging, and land use/land cover (LULC) classification. Due to the significant amount of spectral information from recent HSI sensors, analyzing acquired images challenging using traditional Machine Learning (ML) methods. As number frequency bands increases, required training samples increases exponentially achieve a reasonable accuracy, also known curse...
Coronavirus disease 2019 (COVID-19) has emerged the need for computer-aided diagnosis with automatic, accurate, and fast algorithms. Recent studies have applied Machine Learning algorithms COVID-19 over chest X-ray (CXR) images. However, data scarcity in these prevents a reliable evaluation potential of overfitting limits performance deep networks. Moreover, networks can discriminate pneumonia usually from healthy subjects only or occasionally, limited types. Thus, there is robust accurate...
The band selection in the hyperspectral image (HSI) data processing is an important task considering its effect on computational complexity and accuracy. In this work, we propose a novel framework for problem: Self-Representation Learning (SRL) with Sparse 1D-Operational Autoencoder (SOA). proposed SLR-SOA approach introduces autoencoder model, SOA, that designed to learn representation domain where are sparsely represented. Moreover, network composes of 1D-operational layers non-linear...
Support estimation (SE) of a sparse signal refers to finding the location indices nonzero elements in representation. Most traditional approaches dealing with SE problems are iterative algorithms based on greedy methods or optimization techniques. Indeed, vast majority them use recovery (SR) techniques obtain support sets instead directly mapping locations from denser measurements (e.g., compressively sensed measurements). This study proposes novel approach for learning such training set. To...
Restoration of poor-quality medical images with a blended set artifacts plays vital role in reliable diagnosis. As pioneer study blind X-ray restoration, we propose joint model for generic image restoration and classification: Restore-to-Classify Generative Adversarial Networks (R2C-GANs). This is the first approach forming an Image-to-Image translation task from having noisy, blurry, or over/under-exposed to high-quality domain where forward inverse transformations are learned using...
In this study, the most commonly used polarimetric SAR features including complete coherency (or covariance) matrix information, obtained from several coherent and incoherent target decompositions, backscattering power visual texture are compared in terms of their classification performance different terrain classes. For pattern recognition, two powerful machine learning techniques, Collective Network Binary Classifier (CNBC) with incremental training capability Support Vector Machines (SVM)...
In this work, we propose a novel classification approach based on dual-band one-dimensional Convolutional Neural Networks (1D-CNNs) for of multifrequency polarimetric SAR (PolSAR) data. The proposed can jointly learn from C- and L-band data improve the single band accuracy. To best our knowledge, is first study that introduces 1D-CNNs to land use/land cover domain using PolSAR aims achieve maximum accuracy by one-time training over multiple frequency bands with limited labelled Moreover,...
The frequent breakdowns and malfunctions of industrial equipment have driven increasing interest in utilizing cost-effective easy-to-deploy sensors, such as microphones, for effective condition monitoring machinery. Microphones offer a low-cost alternative to widely used sensors with their high bandwidth capability detect subtle anomalies that other might less sensitivity. In this study, we investigate malfunctioning machines evaluate compare anomaly detection performance across different...
In this work, we propose to use learned features for terrain classification of Polarimetric Synthetic Aperture Radar (PolSAR) images. the proposed framework, are extracted from sliding window regions using Convolutional Neural Networks (CNNs), and then they used with linear Support Vector Machine (SVM) classifier. The performance approach is compared numerous target decomposition theorems (TDs) as engineered tested two classifiers: Collective Network Binary Classifiers (CNBCs) SVMs....
Deep learning-based informative band selection methods on hyperspectral images (HSI) recently have gained intense attention to eliminate spectral correlation and redundancies. However, the existing deep either need additional post-processing strategies select descriptive bands or optimize model indirectly, due parameterization inability of discrete variables for procedure. To overcome these limitations, this work proposes a novel end-to-end network selection. The proposed is inspired by...