An Efficient Multispectral Image Classification and Optimization Using Remote Sensing Data

Journal of Sensors, Journal Year: 2022, Volume and Issue: 2022, P. 1 - 11

Published: Oct. 8, 2022

A significant amount of effort and cost is required to collect training samples for remote sensing image classifications. The study how read multispectral images becoming more important. High-dimensional are created by the various bands that show materials behave. need information about things improvement sensor resolutions have led creation data with a higher size. In recent years, it has been shown high dimensionality these makes hard preprocess them in multiple ways. Recent research demonstrated one most crucial methods address this issue adopting variety learning strategies. But as gets complicated, methodologies not adequate support. proposed methodology shows classification experiment using indicates maximum likelihood classifier different deep models; weight vector (WV) AdaBoost ADAM can greatly limit overfitting, obtains accuracy. Proposed VGG16 Inception v3 increase accuracy along optimization process produce 96.08%.

Language: Английский

---

Evaluation of Human Pose Recognition and Object Detection Technologies and Architecture for Situation-Aware Robotics Applications in Edge Computing Environment

IEEE Access, Journal Year: 2023, Volume and Issue: 11, P. 92735 - 92751

Published: Jan. 1, 2023

Realization of situation-awareness for autonomous robotics applications in edge computing environment is challenging.First, capabilities devices are limited, which must be considered the execution machine learning (ML)-based solutions.Second, many technologies available realizing situation-aware capabilities, but comparison and integration solutions creates additional challenges.Third, existing ML-based models often not directly applicable custom applications, model(s) may need to re-trained with new data.The contribution this paper efficiency feasibility evaluation human pose recognition object detection environment.Several lessons learnt covering constraints presented regarding experimented data sets.The results indicated that simultaneous (Google's Movenet) (Yolov5) on Jetson AGX Xavier achieved ~13-16 FPS, while GPU CPU utilization remained at a medium level, most memory unused (< 44 %).Object concept activation algorithms as an contribution.Realized architecture design prototyped system multiple environments can partial big reference architecture.

Language: Английский

---

Medicine Drug Name Detection Based Object Recognition Using Augmented Reality

Frontiers in Public Health, Journal Year: 2022, Volume and Issue: 10

Published: April 29, 2022

Augmented Reality (AR) is an innovation that empowers us in coordinating computerized data into the client's real-world space. It offers advanced and progressive methodology for medicines, providing medication training. AR aids surgery planning, patient therapy discloses complex medical circumstances to patients their family members. With accelerated upgrades innovation, ever-increasing number of records get accessible, which contain a lot sensitive data, similar substances relations between them. To exploit clinical texts these it important separate significant from texts. Drugs, along with some kind fundamental components, additionally should be perceived. Drug name recognition (DNR) tries recognize drugs specified unstructured order them predefined classifications, utilized deliver connected 3D model inside present reality client This work shows utilization give active visual representation about medicines applications. The proposed method mobile application uses native camera optical character algorithm (OCR) extract text on medicines. extracted over above processed using natural language processing (NLP) tools are then used identify generic category drug dedicated DNR database. database system scraped various resources studies named medi-drug development standpoint. prepared particularly presented ArCore. results obtained encouraging. can detect average time 0.005 s produce output 1.5 s.

Language: Английский

---

An Optimized Hyperparameter of Convolutional Neural Network Algorithm for Bug Severity Prediction in Alzheimer’s-Based IoT System

Computational Intelligence and Neuroscience, Journal Year: 2022, Volume and Issue: 2022, P. 1 - 14

Published: June 28, 2022

Softwares are involved in all aspects of healthcare, such as booking appointments to software systems that used for treatment and care patients. Many vendors consultants develop high quality healthcare hospital management systems, medical electronic middle-ware softwares devices. Internet Things (IoT) devices gaining attention facilitate the people with new technology. The health condition patients monitored by IoT using sensors, specifically brain diseases Alzheimer, Parkinson's, Traumatic injury. Embedded is present complexity increases day-by-day increase number bugs Bugs can have severe consequences inaccurate records, circulatory suffering, death some cases along delay handling There a need predict impact (severe or nonsevere), especially case due their critical nature. This research proposes hybrid bug severity prediction model convolution neural network (CNN) Harris Hawk optimization (HHO) based on an optimized hyperparameter CNN HHO. dataset created, consists devices, which evaluation proposed model. A preprocessing technique textual applied feature extraction embedding layer. In HHO, we define values "Batch Size, Learning Rate, Activation Function, Optimizer Parameters, Kernel Initializers," before training Hybrid CNN-HHO applied, 10-fold cross validation performed evaluation. Results indicate accuracy 96.21%

Language: Английский

---

A two-stage surrogate-assisted meta-heuristic algorithm for high-dimensional expensive problems

Soft Computing, Journal Year: 2023, Volume and Issue: 27(10), P. 6465 - 6486

Published: Feb. 1, 2023

Language: Английский

---