Implementation of YOLO8 for real-time object recognition and tracking for visually impaired

AIP conference proceedings, Journal Year: 2025, Volume and Issue: 3157, P. 070014 - 070014

Published: Jan. 1, 2025

Language: Английский

---

Hybrid Machine Learning Approaches for Predicting and Diagnosing Major Depressive Disorder

International Journal of Advanced Computer Science and Applications, Journal Year: 2024, Volume and Issue: 15(3)

Published: Jan. 1, 2024

Major Depressive Disorder (MDD) is common and debilitating, requiring accurate prediction diagnosis. This study uses clinical, demographic, EEG data to test hybrid machine learning methods for MDD reveals brain electrical activity can identify patterns traits. The aimed enhance diagnosis using methods, focusing on alongside clinical demographic information. Employing various algorithms like CatBoost, Random Forest, XG Boost, XGB SVM with a linear kernel, logistic regression Elasticnet regularization, the found that CatBoost achieved highest accuracy of 93.1% in diagnosis, surpassing other models. Additionally, ensemble model combining XGBoost Forest showed strong performance ROC analysis, effectively discriminating between individuals without MDD. These findings underscore potential integration techniques accurately identifying classifying patients, paving way personalized interventions targeted treatments depressive disorders.

Language: Английский

---

Discovery of astronomical objects in galaxies by means of deep learning

MATEC Web of Conferences, Journal Year: 2024, Volume and Issue: 392, P. 01123 - 01123

Published: Jan. 1, 2024

The study of space exploration has been studied for a very long time, and as technology developed, so too have the methods techniques employed, along with quantity type data acquired. We now receive much astronomical data, brand new is being generated every day, that it physically impracticable to examine all only by human work. In our study, we look at number astronomers face while working this massive amount they use deep learning discover best each objective. [1]previously SVM, KNN, random forest approach, decision trees, other multi-class classification algorithms are used in methodology. Galaxies' propensity belong specific classes forecasted using regression. even if findings from method were was unable effectively divide galaxies into five groups. This approach does not explain real-time categorization take outliers consideration. model's adaptability constrained. scheme account modelling well their evolution. Here, suggest Inception v3 VGG-19 image analysis. Segmentation discovering classifying galaxies. These greatly advance certain fields where there enormous volumes duplicate must be deleted accordance demands thanks Python's high performance investigation processing computer vision.. As result, less need researchers carefully sort through collected satellite telescopes, sky surveys, etc. [2]

Language: Английский

---

The Implementation of Transfer Learning by Convolution Neural Network (CNN) for Recognizing Facial Emotions

Journal of Advanced Research in Applied Sciences and Engineering Technology, Journal Year: 2023, Volume and Issue: 32(2), P. 255 - 276

Published: Sept. 7, 2023

The primary objective of this study is to develop a real-time system that can predict the emotional states an individual who commonly undergoes various experiences. methodology suggested in research for detecting facial expressions involves integration transfer learning techniques incorporate convolutional neural networks (CNNs), along with parameterization approach minimizes number parameters. FER-2013, JAFFE, and CK+ datasets were jointly used train CNN architecture detection, which broadened range may be recognized. proposed model has capability identify emotions, including but not limited happiness, fear, surprise, anger, contempt, sadness, neutrality. Several methods employed assess efficacy model's performance study. experimental results indicate surpasses previous studies terms both speed accuracy.

Language: Английский

---

Detection and Evaluation of Cervical Cancer by Multiple Instance Learning

Published: July 19, 2023

In terms of frequency, cervical cancer is the fourth most common cause cancer-related deaths in women worldwide. Early diagnosis intraepithelial neoplasia (CIN) can significantly improve patient survival chances. This study intends to develop a cutting-edge deep learning architecture perform accurate CIN recognition. Here, ResNet50-based Convolution Neural Network (CNN)-based transfer technique used. The dataset, assembled using some images from internet and that have been enhanced trained CNN. CNN performs well correctly forecasts whether disease will be favorable or negative.

Language: Английский

---

Image encryption using enhanced DCT transform for frequency domain applications

MATEC Web of Conferences, Journal Year: 2024, Volume and Issue: 392, P. 01121 - 01121

Published: Jan. 1, 2024

improving system performance and reliability. Key leaks inadequate key strength are common issues with numerical global models public block cipher algorithms that use pseudo-random numbers. Additionally, because of the spatial correlation between neighboring pixel values, no current crypto scheme is appropriate for picture encryption. An intellectual domain-based image cryptosystem hierarchical modification structures proposed to overcome all these obstacles. The suggested framework also aims maximize confusion metrics improve security efficiency. In this paper modified DCT transform domain based frequency encryption along optimized grouping formulation reduced complexity overhead. model can achieve 100 percentage transformations irrespective non-linear characteristics associated dimensions. To validate different biomedical images collected superiority verified using quantitative qualitative analyzes. implementation helps in offering privacy network security. It validated hybrid combination extend applicability EMR applications.

Language: Английский

---

Reducing Loss for Brain Tumour Detection and Classification in MRI using Deep Learning Techniques

Deleted Journal, Journal Year: 2024, Volume and Issue: 31(6s), P. 330 - 341

Published: Aug. 15, 2024

The signs of a brain tumour might be general or specific. typical symptom is brought on by pressing against the spinal cord. When has impacted certain section and that area not working correctly. Identification tumours was extremely difficult complicated due to location, kind, shape, size in brain. Brain diagnosis since tumor's resolution cannot clearly evaluated early stages growth. However, if found diagnosed stage growth, there good chance patient will treated. As result, essential for successful treatment. To solve this problem, we suggested an automated method can more help detect classify three different types MRI using deep learning techniques, faster RCNN+ResNet50.

Language: Английский

---

An Efficient License Plate Number Recognition System for Traffic Surveillance Using Deep Neural Networks

Lecture notes in electrical engineering, Journal Year: 2024, Volume and Issue: unknown, P. 1032 - 1037

Published: Oct. 19, 2024

Language: Английский

---

Usage of Computer Aided Methods for Detection and Evaluation of Breast Cancer in Mammograms

Published: Sept. 18, 2024

Language: Английский

---

Application of Novel Deep Learning Techniques for Brain Stroke Prediction

Lecture notes in electrical engineering, Journal Year: 2024, Volume and Issue: unknown, P. 1059 - 1064

Published: Oct. 19, 2024

Language: Английский

---