Digital twins in precision agriculture monitoring using artificial intelligence

Elsevier eBooks, Journal Year: 2023, Volume and Issue: unknown, P. 243 - 265

Published: Jan. 1, 2023

Language: Английский

---

Medical Image Encryption using Biometric Image Texture Fusion

Journal of Medical Systems, Journal Year: 2023, Volume and Issue: 47(1)

Published: Nov. 4, 2023

Language: Английский

---

Deep learning-based Object Detection in Underwater Communications System

Published: April 5, 2023

Being at the nexus of robotics and ocean engineering, underwater robots have been a developing research area. They can be used for deep sea infrastructure inspections, oceanographic mapping, environmental monitoring. Autonomous navigation skills are essential doing these activities successfully, especially given poor communication conditions in locations. technologies, such as path planning tracking, one fascinating but difficult issues field study due to extremely dynamic three-dimensional settings. Due their short detection ranges visibility, cameras not received much attention an sensor. However, using visual data from is still popular technique sensing, it works particularly well close-range detections. In this study, enhancement vision achieved by combining max-RGB shades grey methods. Then, solve problem poorly illuminated images, known RCNN (Region-based Convolutional Neural Network) proposed. This procedure tells mapping relationship how create illumination map. Following image processing, strategy classification recommended. Two improved strategies then change structure accordance with properties vision. order deal challenges object tracking communication, correlation filter algorithm (CFTA) method was created. The invariant moment area were looked after object's region had extracted threshold segment morphological technique. findings show that suggested effective target based on RCNN-CFTA aquatic environment. Simulated evaluation methods' performance demonstrates potency strategies.

Language: Английский

---

Applications of Deep Learning in Healthcare in the Framework of Industry 5.0

Advances in web technologies and engineering book series, Journal Year: 2024, Volume and Issue: unknown, P. 69 - 85

Published: Jan. 25, 2024

Emergence of deep learning (DL) and its applicability motivated researchers scientists to explore applications in their fields expertise. In medical technology, a huge amount data is required, dealing with challenging task for researchers. The emergence neural networks modifications like convolutional (CNN), generative adversarial network (AGN), recurrent (RNN), subcategories has provided stage flourish learning. DL been successful tool the pattern recognition, natural language processing (NLP), image processing, speech computer vision, etc. All these techniques have employed healthcare. Image proven be fruitful technique physicians properly diagnose patients through CT scan, MRI, PET, radiography, nuclear medicine, ultrasound, this chapter, some healthcare envisaged, it concluded that very

Language: Английский

---

Automatic Whitefly Detection Algorithm Using Image Segmentation and Feature Analysis in Plant Leaf

Published: April 4, 2024

Language: Английский

---

Towards accurate diagnosis: exploring knowledge distillation and self-attention in multimodal medical image fusion

Journal of Experimental & Theoretical Artificial Intelligence, Journal Year: 2024, Volume and Issue: unknown, P. 1 - 30

Published: Oct. 13, 2024

Multimodal medical image fusion aims to aggregate significant information based on the characteristics of images from different modalities. Existing research in faces several major limitations, including a scarcity paired data, noisy and inconsistent modalities, lack contextual relationships, suboptimal feature extraction techniques. In response these challenges, this proposes novel adaptive approach. Our knowledge distillation (KD) model extracts informative features multimodal using various key components. A teacher network is employed emphasise suitability complexity capturing high-level abstract features. The soft labels are utilised transfer between as well student network. During training, we minimise divergence labels. To enhance extracted apply self-attention mechanism. Training mechanism minimises loss function, encouraging attention scores capture relevant relationships Additionally, cross-modal consistency module aligns ensure spatial meaningful fusion. strategy effectively combines diagnostic value quality fused images. We employ generator discriminator architectures for synthesising distinguishing real generated Comprehensive analysis conducted basis diverse evaluation measures. Experimental results demonstrate improved outcomes with values 0.92, 41.58, 7.25, 0.958, 0.759, 0.947, 0.90, 7.05, 0.0726, 76 s SSIM, PSNR, FF, VIF, UIQI, FMI, EITF, entropy, RMSE, execution time, respectively.

Language: Английский

---

An Effective Framework for Detecting Epileptic Seizures using CNN and Encrypted EEG Signals

Published: June 14, 2023

Electroencephalogram (EEG) signals may be used to autonomously diagnose epilepsy, eliminating the requirement for a medical professional's involvement in process. A good classification performance is really necessary if you do not want pass up any possible discoveries. In this piece of research, authors offer technique automated identification epilepsy using EEG waves. order extract features, raw were first put through discrete Fourier transform, also known as DFT, well wavelet transform (DWT). Therefore, researchers have been exploring different methods improve accuracy signal analysis. One such method proposed study Wavelet Transform based Fourier-Bessel series expansion (WT - FBSE) method. This utilizes WT FBSE spectrum segment multiple frame-size time-segmented scale-space boundary detection The decomposes into narrow sub-band signals, which are then various features log-energy-entropy (LEnt), line-length (LL), and norm-entropy (NEnt) from frequency ranges. choose that most important, relief-F feature ranking approach applied. helps limit amount computing work required by models. evaluates two time-segmentation approaches four frame sizes analyse achieves better when compared existing systems.

Language: Английский

---

Integrated Model for Covid 19 Disease Diagnosis using Deep Learning Approach

Published: July 19, 2023

The severe Corona Virus Disease-2019 (COVID-19), which is caused by the acute respiratory syndrome-Corona Virus-2 (SARS-CoV-2), has killed millions of people worldwide. Imaging methods like Chest X-rays (CXR) and Computed Tomography (CT) are frequently utilised to diagnose COVID-19 quickly reliably. However, manual identification infections through radiographic imaging challenging, time-consuming, prone human error. Deep learning, particularly Convolutional Neural Networks (CNN), preferred approach for identifying extracting features from such medical images. This study employs CNN differentiate between healthy lungs, lungs affected COVID-19, viral pneumonia-affected lungs. ultimate goal categorizing data develop a model or tool that can distinguish various diseases individuals predict disease status. To address these challenges, new network named ResNet50 developed lung CT segmentation with aim creating an effective neural uses fewer training

Language: Английский

---

A Novel System Model for Managing Cyber Threat Intelligence

2022 IEEE 2nd Mysore Sub Section International Conference (MysuruCon), Journal Year: 2022, Volume and Issue: unknown, P. 1 - 5

Published: Oct. 16, 2022

Cyber threat intelligence (CTI) systems, which collect CTI statistics since publically accessible resources, have been the subject of much study as a means mitigating ever-evolving cyber dangers. Due to ever-increasing sophistication and persistence attackers, well lightning-fast pace at assaults develop, quick decision making is now crucial sustained security most companies. As result, several businesses started using management systems better coordinate their defences against threats those other Getting handle on how these platforms should be built, deployed, utilised requires first knowing successful they in past. However, lack consensus what aspects affect performance exists between academia industry. We used review entrenched methodology gather data from 152 experts order empirically evaluate concerns. Then, we determined few variables that are critical effectiveness platform inside an organisation.

Language: Английский

---

Brain Tumor Detection using Machine Learning Techniques with Internet of Things

2022 IEEE 2nd Mysore Sub Section International Conference (MysuruCon), Journal Year: 2022, Volume and Issue: unknown, P. 1 - 7

Published: Oct. 16, 2022

Epilepsy is a neurological condition that rather common and thought to afflict around 70 million individuals all over the globe. If epilepsy be monitored properly successfully treated, seizures have recorded logged. The present therapy for involves use of seizure diaries kept by caregivers; nevertheless, clinical detection may sometimes miss events. Wearable technologies may, in long term, prove less intrusive, more pleasant, simpler ambulatory monitoring. Using biosensors placed on wrist ankle, custom-built machine learning (ML) algorithms are tested see whether or not they able correctly recognise broad spectrum epileptic episodes. In this article, an automated method known as new wireless sensor-based system developed purpose detecting monitoring patients environment setting. goal technique cut down amount time spent neurologists diagnosing seizures. biosensor worn wrist, was devised study recording multi-modal data such electroencephalogram (EEG) readings. However, excluding noise extracting features two important challenges must overcome when attempting foresee Support Vector Machine (SVM) used classifier obtain statistical values Lyapunov than raw order detect activity shorter time. This resulted significant improvement compared methods currently considered state-of-the-art.

Language: Английский

---