convolutional neural network
در نشریات گروه برق-
Deep Learning Approach for Forest Fire Detection: A CNN Classification Model on the DeepFire Dataset
Forests play several vital roles in our lives and provide various resources. However, in recent years, the increasing frequency of wildfires has led to the widespread burning and destruction of many forests and wildlands. Therefore, detecting forest fires and finding suitable solutions to address this issue has become one of the critical challenges for researchers. Today, with the advancement of artificial intelligence, forest fire detection using deep learning is an important method with the aim of increasing the efficiency of forest fire detection and monitoring systems. In this article, a method based on a type of convolutional neural network called Xception is proposed for classifying forest fire images. In this method, transfer learning technique is used on the proposed neural network and a new classifier is designed for the problem. Also, various hyperparameters have been used to optimize the performance of the proposed model. The proposed method is performed on the DeepFire dataset, which contains 1900 images equally divided between fire and no-fire classes. The results obtained from the implementation of the proposed method show that this method with an accuracy of 99.47% has achieved a favorable performance in classifying forest fire images.
Keywords: Artificial Intelligence, Classification, Convolutional Neural Network, Deep Learning, Forest Fire Detection -
This paper introduces a high-accuracy prediction framework for Medium Term Load Forecasting (MTLF) in a manufacturing plant at Iran Khodro Company, leveraging the power of deep learning. Specifically, the proposed method integrates a Convolutional Neural Network (CNN) with a Long Short-Term Memory (LSTM) neural network, effectively capturing both spatial and temporal dependencies in the data. The performance of this hybrid deep learning model is rigorously evaluated against traditional regression techniques, including Linear Regression, Ridge Regression, and Lasso Regression. experimental results demonstrate a remarkable coefficient of determination (R² score) of 0.95 on the test dataset when employing the deep neural network model, significantly outperforming classical methods, which achieve an R² score of only 0.81. This substantial improvement underscores the superior predictive accuracy and generalization capability of the proposed approach. the model is trained using historical energy consumption data, where past electricity load values serve as inputs for the deep learning architecture. The dataset consists of nine years of monthly energy consumption records (2011–2019) collected from Iran Khodro Company, providing a robust foundation for medium-term load forecasting. The findings of this study highlight the effectiveness of deep learning in industrial energy demand prediction, offering a reliable and scalable solution for optimizing energy management in manufacturing sectors.
Keywords: Convolutional Neural Network, Deep Neural Network, Linear Regression, Long Short-Term Memory Neural Network (LSTM), Medium Term Load Forecasting (MTLF) -
Artificial intelligence (AI) has significantly advanced speech recognition applications. However, many existing neural network-based methods struggle with noise, reducing accuracy in real-world environments. This study addresses isolated spoken Persian digit recognition (zero to nine) under noisy conditions, particularly for phonetically similar numbers. A hybrid model combining residual convolutional neural networks and bidirectional gated recurrent units (BiGRU) is proposed, utilizing word units instead of phoneme units for speaker-independent recognition. The FARSDIGIT1 dataset, augmented with various approaches, is processed using Mel-Frequency Cepstral Coefficients (MFCC) for feature extraction. Experimental results demonstrate the model’s effectiveness, achieving 98.53%, 96.10%, and 95.92% accuracy on training, validation, and test sets, respectively. In noisy conditions, the proposed approach improves recognition by 26.88% over phoneme unit-based LSTM models and surpasses the Mel-scale Two Dimension Root Cepstrum Coefficients (MTDRCC) feature extraction technique along with MLP model (MTDRCC+MLP) by 7.61%.
Keywords: Spoken Digit Recognition, Data Augmentation, Convolutional Neural Network, Bidirectional Gated Recurrent Unit -
Journal of Electrical and Computer Engineering Innovations, Volume:13 Issue: 2, Summer-Autumn 2025, PP 267 -274Background and ObjectivesPerson re-identification is an important application in computer vision, enabling the recognition of individuals across non-overlapping camera views. However, the large number of pedestrians with varying appearances, poses, and environmental conditions makes this task particularly challenging. To address these challenges, various learning approaches have been employed. Achieving a balance between speed and accuracy is a key focus of this research. Recently introduced transformer-based models have made significant strides in machine vision, though they have limitations in terms of time and input data. This research aims to balance these models by reducing the input information, focusing attention solely on features extracted from a convolutional neural network model.MethodsThis research integrates convolutional neural network (CNN) and Transformer architectures. A CNN extracts important features of a person in an image, and these features are then processed by the attention mechanism in a Transformer model. The primary objective of this work is to enhance computational speed and accuracy in Transformer architectures.ResultsThe results obtained demonstrate an improvement in the performance of the architectures under consistent conditions. In summary, for the Market-1501 dataset, the mAP metric increased from approximately 30% in the downsized Transformer model to around 74% after applying the desired modifications. Similarly, the Rank-1 metric improved from 48% to approximately 89%.ConclusionIndeed, although it still has limitations compared to larger Transformer models, the downsized Transformer architecture has proven to be much more computationally efficient. Applying similar modifications to larger models could also yield positive effects. Balancing computational costs while improving detection accuracy remains a relative goal, dependent on specific domains and priorities. Choosing the appropriate method may emphasize one aspect over another.Keywords: Person Re-Identification, Deep Learning, Image Processing, Convolutional Neural Network, Computer Vision
-
در این مقاله یک سیستم تشخیص خودکار موارد مبتلا به کوید-19 مبتنی بر اینترنت اشیا پیشنهاد می شود. در مدل پیشنهادی ابتدا با استفاده از فن آوری اینترنت اشیا تصاویر پزشکی مستقیم پس از مراجعه فرد مشکوک از طریق تجهیزات پزشکی مجهز به اینترنت اشیا به مخزن داده ارسال می شود. سپس به منظور کمک به متخصصین رادیولوژی برای تفسیر هرچه بهتر تصاویر پزشکی از چهار مدل شبکه عصبی پیچشی از پیش آموزش دیده به نام های InceptionResNetV2، InceptionV3، VGG19 و ResNet152 و دو مجموعه داده تصاویر پزشکی رایولوژی قفسه سینه و CT Scan در یک طبقه بندی سه کلاسه برای پیش بینی دقیق موارد مبتلا به کوید-19، افراد سالم و موارد مبتلا بیماری استفاده می شود. درنهایت بهترین نتیجه به دست آمده برای تصاویر CT Scan متعلق به معماری InceptionResNetV2 با دقت 99.366% و برای تصاویر رادیولوژی مربوط به معماری InceptionV3 با دقت 96.943% می باشد. نتایج نشان می دهد این سیستم منجر به کاهش مراجعه روزانه به مراکز درمانی و در نتیجه کاهش فشار بر سیستم مراقبت های درمانی می شود. همچنین به متخصصین رایولوژی و کادر درمان کمک می کند تا هرچه سریعتر بیماری شناسایی شود.
کلید واژگان: پردازش تصویر، هوش مصنوعی، اینترنت اشیا، شبکه عصبی پیچشی، یادگیری عمیقIn this paper, we propose an automatic detection system for COVID-19 cases based on the Internet of Things. In the proposed model, first, using Internet of Things technology, medical images are sent directly to the data collection after the suspicious person's visit through medical equipment equipped with Internet of Things, and then, in order to help radiologists to interpret medical images better, usage has been made of four pre-trained convolutional neural network models i.e. InceptionV3, InceptionResNetV2, VGG19 and ResNet152 as well as two datasets of chest radiology medical images and CT Scan in a 3-class classification for accurate prediction of cases suffering from COVID-19, healthy people, and diseased cases. Finally, the best result for CT-Scan images is related to InceptionResNetV2 architecture with an accuracy of 99.366%, and for radiology images related to the InceptionV3 architecture, it is 96.943%. The results show that this system leads to a reduction in daily visits to medical centers and thus reduces the pressure on the medical care system. It also helps rheology specialists to identify the disease as quickly as possible.
Keywords: Image Processing, Artificial Intelligence, Internet Of Things, Convolutional Neural Network, Deep Learning -
Accurate prediction of stock market trends is crucial for informed investment decisions and effective portfolio management, ultimately leading to enhanced wealth creation and risk mitigation. This study proposes a novel approach for predicting stock prices in the stock market by integrating Convolutional Neural Networks (CNN) and Long Short-Term Memory (LSTM) networks, using sentiment analysis of social network data and candlestick data (price). The proposed methodology consists of two primary components: sentiment analysis of social networks and candlestick data. By amalgamating candlestick data with insights gleaned from Twitter, this approach facilitates a more detailed and accurate examination of market trends and patterns, ultimately leading to more effective stock price predictions. Additionally, a Random Forest algorithm is used to classify tweets as either positive or negative, allowing for a more subtle and informed assessment of market sentiment. This study uses CNN and LSTM networks to predict stock prices. The CNN extracts short-term features, while the LSTM models long-term dependencies. The integration of both networks enables a more comprehensive analysis of market trends and patterns, leading to more accurate stock price predictions.
Keywords: Stock Price Prediction, Deep Learning, Sentiment Analysis, Long Short-Term Memory, Convolutional Neural Network -
Scientia Iranica, Volume:31 Issue: 21, Nov-Dec 2024, PP 1939 -1947Plant diseases are a signi cant concern in agriculture, contributing to as much as 16% of global agricultural losses. This poses serious threats to food security, especially for crops like bananas, which are highly vulnerable to diseases such as Xanthomonas Wilt and Sigatoka leaf spot. These diseases have the potential to cause complete yield losses, reaching up to 100%. Addressing these challenges is crucial, and this study aims to do so by developing a robust disease detection model. Leveraging Convolutional Neural Network (CNN) algorithms, we have created a sophisticated system capable of accurately identifying and categorizing diseases in banana plants. To train our model e ectively, we have gathered a meticulously curated dataset of banana plant leaf images from regions heavily a ected by these diseases. This dataset has been carefully categorized into three groups: Healthy, Xanthomonas Wilt infected, and Sigatoka leaf spot infected. Employing advanced techniques such as data augmentation and transfer learning, we have netuned our model using various architectures including MobileNet, EcientNet, VGG16, VGG19, and InceptionV3. Our research ndings highlight the exceptional performance of the VGG16 model, achieving an impressive classi cation accuracy of 81.53% during rigoroustesting with independent datasets. Looking to the future, we recognize the need for further improvements in model performance. This includes acquiring a more diverse and expansive dataset and implementing automatic hyperparameter selection methods.Keywords: Convolutional Neural Network, Deep Learning, Plant Disease Detection, Xanthomonas Wilt, Sigatoka Leaf Spot, SDG
-
نشریه پژوهش های نظری و کاربردی هوش ماشینی، سال دوم شماره 2 (پیاپی 4، پاییز و زمستان 1403)، صص 31 -48این مقاله به بررسی کارایی طبقه بندهای ترکیبی CNN-DRNN در شناسایی آپنه خواب با استفاده از سیگنال الکتروکاردیوگرام قلب (ECG) پرداخته است. در این مطالعه، مدل های مختلف شبکه های عصبی کانولوشنی ازجمله AlexNet، VGG16، VGG19 و ZFNet در ترکیب با مدل های شبکه عصبی بازگشتی عمیق شامل LSTM، GRU و BiLSTM مورد ارزیابی قرارگرفته است. این مدل ها با و بدون استفاده از بهینه سازهای هوش جمعی گورکن عسل و گرگ خاکستری برای تعیین مقادیر بهینه ابرپارامترها مقایسه شده اند. نتایج نشان می دهد که مدل ترکیبی AlexNet-GRU پس از اعمال هر دو بهینه ساز، بهترین عملکرد را با دقت 95٪، نرخ تشخیص 61/97٪ و F-Score 37/93٪ ارائه کرده است. در این پژوهش، چالش بهینه سازی ابرپارامترها در مدل های یادگیری عمیق با استفاده از دو بهینه ساز گورکن عسل و گرگ خاکستری بررسی شده است. این بهینه سازها با الهام از رفتارهای طبیعت، تعامل غیرمستقیم میان عامل ها و توزیع هوشمند به حل این چالش کمک می کنند. البته، بهینه ساز گورکن عسل در مقایسه با گرگ خاکستری در انتخاب مقادیر بهینه ابرپارامترها عملکرد بهتری از خود نشان داده است.کلید واژگان: آپنه خواب، بهینه ساز گورکن عسل، بهینه ساز گرگ خاکستری، شبکه عصبی پیچشی، شبکه عصبی بازگشتی عمیقJournal of Applied and Basic Machine Intelligence Research, Volume:2 Issue: 2, Autumn & Winter 2024, PP 31 -48This study investigates the efficiency of CNN-DRNN hybrid classifiers in detecting sleep apnea using electrocardiogram (ECG) signals. Various CNN models were evaluated, including AlexNet, VGG16, VGG19, and ZFNet, along with DRNN models such as LSTM, GRU, and BiLSTM. These models were compared with and without the application of swarm intelligence optimizers, namely the Honey Badger Algorithm (HBA) and Grey Wolf Optimizer (GWO), for optimizing hyperparameter values. The results demonstrated that the AlexNet-GRU hybrid model achieved the best performance after applying both optimizers, with an accuracy of 95%, a detection rate of 97.61%, and an F-Score of 93.37%.This research also explores the challenges of hyperparameter optimization in deep learning models using swarm intelligence-based optimizers. These optimizers, inspired by natural behaviors, facilitate problem-solving through intelligent distribution, indirect interactions among agents, and simplification of complex processes. Additionally, the findings revealed that HBA outperformed GWO in determining optimal hyperparameter values, leading to enhanced model performance. Overall, the study highlights the potential of integrating deep learning models with swarm intelligence optimizers to improve sleep apnea detection.Keywords: Sleep Apnea, Honeybadger Optimizer, Greywolf Optimizer, Convolutional Neural Network, Deep Recurrent Neural Network
-
نشریه پژوهش های نظری و کاربردی هوش ماشینی، سال دوم شماره 2 (پیاپی 4، پاییز و زمستان 1403)، صص 163 -184رابط مغز -رایانه مبتنی بر گفتار راهبردهای ارتباط صوتی موثری را برای کنترل دستگاه ها از طریق دستورات گفتاری که از سیگنال های مغزی تفسیر می شوند، ارائه می کنند. یکی از چالش های مهم در مسئله رابط مغز-رایانه رده بندی سیگنال های مغزی مبتنی بر الکتروانسفالوگرافی است. الکتروانسفالوگرافی یک سیگنال مغزی غیرتهاجمی است که از سطح پوست سر از طریق الکترودها ضبط می شود. سیگنال های به دست آمده با استفاده از تجهیزات آسان و ارزان دارای وضوح مکانی نسبتا پایین و وضوح زمانی بالا هستند که برای دستیابی به نتایج بهینه مناسب ترین روش استخراج و رده بندی ویژگی ها باید استفاده شود. همچنین جمع آوری داده های کافی برای آزمودنی جدید زمان و تلاش زیادی می طلبد که در این مقاله افزودن داده ها با استفاده از مدل مولد هماوردانه به منظور بهبود عملکرد رده بندی سیگنال های مغزی پیشنهادشده است. همچنین مدلی برای رده بندی تصور گفتار مبتنی بر سیگنال های مغزی آزمودنی جدید با یادگیری انتقالی با استفاده از روش مولد هماوردانه مبتنی بر تطبیق دامنه متمایزکننده هماوردانه ارائه شده است. به منظور شناسایی تصور گفتار از پایگاه داده KaraOne استفاده شده است. روش پیشنهادی با سایر روش های جدید بر اساس معیارهای صحت و کاپا مورد ارزیابی قرار گرفتند. طبق نتایج به دست آمده روش پیشنهادی با صحت 86 ٪ و 21/60 ٪ به ترتیب تصور کلمات و واج ها را رده بندی می کند. مدل پیشنهادی، مستقل از سیگنال های مغزی هر فرد است که با آموزش مدل بر روی سیگنال های مغزی افزوده شده آزمودنی ها در دامنه منبع می توان به طور موثر سیگنال ها را در دامنه هدف بدون نیاز به داده های برچسب گذاری شده از آزمودنی جدید رده بندی کرد.کلید واژگان: تصور گفتار، افزودن داده، تطبیق دامنه متمایزکننده هماوردانه، یادگیری عمیق، یادگیری انتقالی، شبکه عصبی پیچشی، رابط مغز -رایانهJournal of Applied and Basic Machine Intelligence Research, Volume:2 Issue: 2, Autumn & Winter 2024, PP 163 -184Speech-based brain-computer interfaces provide effective voice communication strategies for controlling devices through spoken commands interpreted from brain signals. One of the major challenges in the brain-computer interface problem is the classification of brain signals based on electroencephalography. Electroencephalography is a non-invasive brain signal that is recorded from the scalp surface through electrodes. The signals obtained using easy and cheap equipment have relatively low spatial resolution and high temporal resolution, which requires the most appropriate feature extraction and classification method to achieve optimal results. Also, Collecting sufficient data for a new subject requires a lot of time and effort, so in this paper, data augmentation using a generative adversarial model is proposed to improve the performance of brain signal classification. Also, a model for classifying speech imagery based on brain signals of a new subject with transfer learning using a generative adversarial method based on adversarial domain adaptation method is presented. In order to identify speech imagery, the KaraOne database was used. The proposed method was evaluated with other new methods based on accuracy and kappa criteria. According to the results, the proposed method classifies word imagery and phonemes with 86% and 60.21% accuracy, respectively. The proposed model is independent of each individual's brain signals, which can be effectively classified in the target domain by training the model on the augmented brain signals of the subjects in the source domain without the need for labeled data from the new subject.Keywords: Speech Imagery, Data Augmentation, Adversarial Discriminative Domain Adaptation, Deep Learning, Transfer Learning, Convolutional Neural Network, Brain-Computer Interface
-
Predicting pedestrians' intentions to cross paths with cars, particularly at intersections and crosswalks, is critical for autonomous systems. While recent studies have showcased the effectiveness of deep learning models based on computer vision in this domain, current models often lack the requisite confidence for integration into autonomous systems, leaving several unresolved issues. One of the fundamental challenges in autonomous systems is accurately predicting whether pedestrians intend to cross the path of a self-driving car. Our proposed model addresses this challenge by employing convolutional neural networks to predict pedestrian crossing intentions based on non-visual input data, including body pose, car velocity, and pedestrian bounding box, across sequential video frames. By logically arranging non-visual features in a 2D matrix format and utilizing an RGB semantic map to aid in comprehending and distinguishing fused features, our model achieves improved accuracy in pedestrian crossing intention prediction compared to previous approaches. Evaluation against the criteria of the JAAD database for pedestrian crossing intention prediction demonstrates significant enhancements over prior studies.Keywords: Pedestrian Crossing Intention Detection, Self-Driving Cars, Body Pose Keypoints, Convolutional Neural Network, Semantic Map
-
Journal of Artificial Intelligence in Electrical Engineering, Volume:13 Issue: 49, Spring 2024, PP 19 -24
Diagnosis of the fracture site is done using CT-Scan images and based on the doctor's visual diagnosis. This work is very time-consuming and depends on the doctor and his expertise. Systemic methods can help doctors and specialists and can detect the fracture area and the fracture surface. In fractures, only the location of the fracture is determined, but if we want to diagnose the area, high expertise and experience is needed, or in some cases, MRI images are needed. Convolutional neural networks are very powerful in diagnosing diseases and medical complications and can diagnose them correctly. The high accuracy and ability of convolutional neural networks has made this method popular among researchers, and its use is becoming more widespread every day. In this method, fracture location and fracture depth were determined using convolutional neural network. In this work, first the fracture site and then the fracture area are determined. In this study, the location of hip fracture was detected with complete accuracy and the fracture area was obtained with 99.68 accuracy and 99.82% sensitivity. The obtained results indicate that the proposed method is a suitable method for fracture detection.
Keywords: Bone Fracture, Convolutional Neural Network, Aria Detection, Pelvis Fracture -
Diagnosing sleep and wakefulness is an important method in diagnosing sleep problems. This work is done by specialists based on the physical examination of biological signals such as EEG, EOG, ECG, EMG, etc. The deep learning method based on convolutional neural network is one of the newest and most important methods of analysis, separation, and diagnosis, which is expanding day by day. In this article, the deep learning-based convolutional neural network is used to extract features from the time-frequency domain of the EEG signal to classify sleep stages. Here, from the EEG signal, the time-frequency image of the signal is calculated based on the spectrogram. Then deep features are extracted using a convolutional neural network with Alexnet architecture with 8th-order fully connected layers. Finally, without changing the nature of the signal, sleep stages are detected with acceptable accuracy. Finally, by using the SVM classifier, sleep stages were classified with acceptable accuracy. An accuracy of 99.6% was obtained for the classification of sleep stages, which indicates the ability of the method to distinguish sleep stages.
Keywords: EEG Signal, Deep Learning, Sleep Stages, Convolutional Neural Network -
Nowadays, medical intelligence detection systems have evolved significantly due to advancements in artificial intelligence, however, they face some challenges. Breast cancer diagnosis and classification is one of the medical intelligence systems. There are a variety of screening techniques available to detect breast cancer such as mammography, magnetic resonance imaging, and ultrasound. This research uses the MIAS mammography image dataset and tries to diagnose and classify benign and malignant masses based on image processing and machine learning techniques. Initially, we apply pre-processing for noise reduction and image enhancement using Quantum Inverse MFT, and then image segmentation with the Social Spider Algorithm. The type of mass is then diagnosed by the Convolutional neural network. The results show that the proposed approach has better performance in comparison to others based on some evaluation criteria such as accuracy of 99.57%, sensitivity of 91%, and specificity of 86%.
Keywords: Breast Cancer, Diagnosis, Classification, Quantum Inverse MFT Algorithm, Social Spider Algorithm, Convolutional Neural Network -
ساختار و عملکرد آنتن ها، پهنای باند، بهره و هدایت مهمترین شاخص های عملکرد بشمار می آیند. برای این منظور خط انتقال همگن دست راست-چپ به دلیل تلفات کم، تغییرات فاز، پهنای باند فرکانس، رزونانس مرتبه صفر و منفی، مینیاتورسازی و ساخت آسان از اهمیت و جایگاه بالایی برخوردار و در طراحی آنتن پهن باند و آرایه ای بسیار مناسب است. ساختار دست راست-چپی در آنتن ها به دلیل تفاوت فاز بخش راست در تکرار آرایه ها و ضخامت لایه دچار تاخیر فاز و در نهایت انحراف الگوی تابشی است. از طرفی مسدود شدن خط انتقال در قسمت چپ باعث محدودیت پهنای باند و افزایش میزان تلفات سیستم می گردد. در این مقاله با کمک یادگیری عمیق نقایص کامپوزیت برطرف و بهینه سازی آنتن آرایه ای را شامل شده است. طراحی خط انتقال آنتن پیشنهادی در محدوده 2 الی 7 گیگاهرتز، فرکانس تشدید بهینه 5/4 گیگاهرتز و الگوریتم عصبی کانولوشن، رزونانس دوگانه و سلف مارپیچی در چهار آرایه بر روی پچ بارگذاری شده است. استفاده از شبکه عصبی پیچشی در خط انتقال چپ، تاخیر فاز سمت راست را جبران و در نهایت تغییرات فاز بهینه و اصلاح الگوی تابشی و اسکن مداوم آرایه های فازی را مقدور می سازد. همچنین با ایجاد شکاف در پچ مایکرواستریپ محدودیت پهنای باند برطرف و تلفات سیستم کاهش می یابد. ابعاد ثانویه نسبت به بعد اولیه با توجه به مدل اصلاح شده هوشمند تا حدود 60 درصد کاهش سایز و مینیاتورسازی صورت می گیرد. نتایج این کامپوزیت ارتقا یافته نشان دهنده افزایش پهنای باند 3/20 و بهره وری الگوی تابش بیش از 96 درصد است. از طرفی ابعاد کوچک، پهنای باند فرکانسی مناسب و طراحی ساده شبکه نیز تامین شده است.
کلید واژگان: آنتن آرایه ای، پچ مایکرواستریپ، شبکه عصبی کانولوشن، کامپوزیت همگن، یادگیری عمیق.Journal of Intelligent Procedures in Electrical Technology, Volume:16 Issue: 63, Autumn 2025, PP 141 -154Antenna structure and performance, bandwidth, gain and guidance are the most important performance indicators. For this purpose, RL homogeneous transmission line is very important due to low loss, phase changes, frequency bandwidth, zero and negative order resonance, miniaturization and easy construction, and is very suitable in the design of broadband and array antennas. The right-left hand structure in the antennas due to the difference in the phase of the right part in the repetition of arrays and the thickness of the layer has phase delay and finally deviation of the radiation pattern. On the other hand, the blockage of the transmission line on the left causes bandwidth restriction and increasing the number of casualties in the system. In this paper, with the help of deep learning (DL), composite defects are solved and optimized arrayed antenna. The proposed antenna transmission line design in the range of 2-7 GHz, optimum resonance frequency of 4.5 GHz and convolution, dual resonance and spiral inductor neural algorithm are loaded onto the patch in four arrays. The use of convolutional neural network (CNN) in the left transmission line compensates for the right phase delay and finally enables optimal phase changes, correction of radiation pattern and continuous scanning of phase arrays. Also, by creating gaps in the microstrip patch, bandwidth limit is removed and the system losses are reduced. Secondary dimensions compared to the primary dimension are reduced to about 60% in size and miniature according to the smart modified model. The results of this improved composite showed an increase in bandwidth of 20.3 and the efficiency of the radiation pattern by more than 96%. On the other hand, small dimensions, appropriate frequency bandwidth and simple network design have been provided.
Keywords: Array Antenna, Convolutional Neural Network, Deep Learning, Homogeneous Composite, Micro-Strip Patch -
شناسایی و ارزیابی لکوسیت ها برای ارزیابی کیفیت سیستم ایمنی انسان مهم است. با این حال، تجزیه و تحلیل اسمیر خون به تخصص پاتولوژیست بستگی دارد. روش دستی برای تجزیه و تحلیل و طبقه بندی گلوبولهای سفید ها پرهزینه و زمان بر است و می تواند منجر به خطا در تشخیص شود. اکثر روش های یادگیری عمیق از مدل های مبتنی بر CNN برای طبقه بندی گلبول های سفید استفاده می کنند. این مقاله استفاده از یک شبکه مبتنی بر ViT را برای طبقه بندی لکوسیت ها در نمونه خون مورد بحث قرار می دهد. مجموعه داده مورد استفاده در این مقاله شامل 352 تصویر با اندازه 320 در 240 است که از طریق روش هایی برای ایجاد یک مجموعه داده متعادل از 12444 تصویر داده افزایی شده است. سپس داده های افزایش یافته برای آموزش معماری مبتنی بر ViT برای طبقه بندی انواع مختلف گلبول های سفید مورد استفاده قرار گرفته است. دراولین مرحله از روش پیشنهادی، یک توکنایزر کانولوشن برای استخراج پچ تصاویر اعمال شده است. این پچ ها فلت شده و به عنوان ورودی برای ساختار مبتنی بر ViT برای شناسایی زیر کلاس ها در مرحله دوم استفاده شده اند. نتایج به دست آمده با استفاده از لوکوویت نشان می دهد صحت شبکه پیشنهادی 99.04 درصد است که نسبت به شبکه های پیشرفته برتری دارد.کلید واژگان: گلبول های سفید، طبقه بندی تصویر، یادگیری عمیق، شبکه عصبی کانولوشن، ترانسفورمر بیناییThe identification and evaluation of leukocytes are important to assess the quality of the human immune system; however, the analysis of blood smears depends on the pathologist’s expertise. The manual method for analyzing and classifying WBCs is costly and time-consuming and can result in errors in detection. Most deep learning methods use CNN-based models for white blood cell classification. This paper discusses the use of a ViT-based network, for the classification of leukocytes (WBCs) in a blood sample. The Dataset used in this paper consists of 352 images with a size of 320x240, which was augmented through techniques to create a balanced dataset of 12444 images. The augmented data was then used to train a ViT-based architecture to classify the different types of WBCs. As the first step of the proposed algorithm, a convolutional tokenizer has been applied for patch extraction of images. These patches have been flattened and have been used as input for a ViT-based structure to recognize the subclasses in the second step. The results obtained using Leukovit show that the accuracy of the proposed network is 99.04% which is outperforming the state-of-the-art networks.Keywords: White Blood Cells, Image Classification, Deep Learning, Convolutional Neural Network, Vision Transformer
-
Rice is one of the most important staple crops in the world and provides millions of people with a significant source of food and income. Problems related to rice classification and quality detection can significantly impact the profitability and sustainability of rice cultivation, which is why the importance of solving these problems cannot be overstated. By improving the classification and quality detection techniques, it can be ensured the safety and quality of rice crops, and improving the productivity and profitability of rice cultivation. However, such techniques are often limited in their ability to accurately classify rice grains due to various factors such as lighting conditions, background, and image quality. To overcome these limitations a deep learning-based classification algorithm is introduced in this paper that combines the power of convolutional neural network (CNN) and long short-term memory (LSTM) networks to better represent the structural content of different types of rice grains. This hybrid model, called CNN-LSTM, combines the benefits of both neural networks to enable more effective and accurate classification of rice grains. Three scenarios are demonstrated in this paper include, CNN, CNN in combination with transfer learning technique, and CNN-LSTM deep model. The performance of the mentioned scenarios is compared with the other deep learning models and dictionary learning-based classifiers. The experimental results demonstrate that the proposed algorithm accurately detects different rice varieties with an impressive accuracy rate of over 99.85%, and 99.18% to identify quality for varying combinations of rice varieties with an average accuracy of 99.18%.
Keywords: Rice Classification, Quality Detection, Convolutional Neural Network, Long Short-Term Memory, Transfer Learning -
نشریه دستاوردهای نوین در برق،کامپیوتر و فناوری، سال چهارم شماره 2 (پیاپی 11، تابستان 1403)، صص 23 -33
بیماری کووید-19 که باعث سندرم حاد تنفسی می شود، یک بیماری مسری و کشنده است که اثرات مخربی بر جامعه و زندگی انسان دارد و به طور قابل توجهی بر اقتصاد جهان تاثیر گذاشته است. حیاتی ترین گام در مبارزه با بیماری کووید-19 تشخیص سریع بیماران مبتلا است. تصاویر سی تی قفسه سینه و کیت های تشخیصی RT-PCR اغلب برای تشخیص بیماری استفاده می شوند. هر دو روش ذکر شده با برخی از مشکلات روبرو هستند، به این ترتیب در پژوهش های اخیر از مدل های یادگیری عمیق برای تشخیص بیماری کووید-19 استفاده شده است. مدل های یادگیری عمیق مدل هایی سریع و دقیق هستند که برای تشخیص این بیماری در نظر گرفته شده اند. روش پیشنهادی در این مقاله، استفاده از شبکه عصبی کانولوشن از پیش آموزش دیده برای تشخیص بیماری کووید-19 بر روی دیتاست سی تی اسکن SARS-COV-2 است. این دیتاست شامل1252 سی تی اسکن مثبت برای عفونت کووید-19 و 1230 سی تی اسکن برای بیماران غیر آلوده به عفونت کووید-19 می باشد. شبکه عصبی کانولوشن از پیش آموزش دیده InceptionResNetV2 در مقایسه با سایر شبکه های از پیش آموزش دیده به نتایج بهتری، از جمله صحت 97.59%، دقت 98.78%، بازیابی 96.41% و میانگین F1 %97.58 دست یافته است.
کلید واژگان: یادگیری انتقالی، بیماری کووید-19، تصاویر سی تی قفسه سینه، شبکه عصبی کانولوشن، یادگیری عمیقJournal of New Achievements in Electrical, Computer and Technology, Volume:4 Issue: 2, 2024, PP 23 -33Covid-19, which causes acute respiratory syndrome, is a contagious and fatal disease that has devastating effects on society and human life, and has significantly affected the world economy. The most critical step in the fight against Covid-19 is the rapid diagnosis of infected patients. Chest CT images and RT-PCR diagnostic kits are often used to diagnose the disease. Both mentioned methods face some problems, thus in recent research, deep learning models have been used to diagnose the disease of Covid-19. Deep learning models are fast and accurate models that are considered to diagnose this disease. The proposed method in this article is to use a pre-trained convolutional neural network to diagnose the disease of Covid-19 on the SARS-COV-2 CT scan dataset. This dataset includes 1252 CT scan images belonging to COVID-19 cases and 1230 CT scan images belonging to healthy cases. The pre-trained convolutional neural network InceptionResNetV2 has achieved better results compared to other pre-trained networks, including 97.59% accuracy, 98.78% precision, 96.41% recall and 97.58% F1-Score.
Keywords: Transfer Learning, Covid-19 Disease, Chest CT Scans, Convolutional Neural Network, Deep Learning -
Using intelligent approaches in diagnosing the COVID-19 disease based on machine learning algorithms (MLAs), as a joint work, has attracted the attention of pattern recognition and medicine experts. Before applying MLAs to the data extracted from infectious diseases, techniques such as RAT and RT-qPCR were used by data mining engineers to diagnose the contagious disease, whose weaknesses include the lack of test kits, the placement of the specialist and the patient pointed at a place and low accuracy. This study introduces a three-stage learning framework including a feature extractor by visual geometry group 16 (VGG16) model to solve the problems caused by the lack of samples, a three-channel convolution layer, and a classifier based on a three-layer neural network. The results showed that the Covid VGG16 (CoVGG16) has an accuracy of 96.37% and 100%, precision of 96.52% and 100%, and recall of 96.30% and 100% for COVID-19 prediction on the test sets of the two datasets (one type of CT-scan-based images and one type of X-ray-oriented ones gathered from Kaggle repositories).
Keywords: COVID-19 Prediction, Convolutional Neural Network, Transfer Learning, Computer Vision, Image Processing -
Journal of Electrical and Computer Engineering Innovations, Volume:12 Issue: 2, Summer-Autumn 2024, PP 401 -408Background and ObjectivesRe-identifying individuals due to its capability to match a person across non-overlapping cameras is a significant application in computer vision. However, it presents a challenging task because of the large number of pedestrians with various poses and appearances appearing at different camera viewpoints. Consequently, various learning approaches have been employed to overcome these challenges. The use of methods that can strike an appropriate balance between speed and accuracy is also a key consideration in this research.MethodsSince one of the key challenges is reducing computational costs, the initial focus is on evaluating various methods. Subsequently, improvements to these methods have been made by adding components to networks that have low computational costs. The most significant of these modifications is the addition of an Image Re-Retrieval Layer (IRL) to the Backbone network to investigate changes in accuracy.ResultsGiven that increasing computational speed is a fundamental goal of this work, the use of MobileNetV2 architecture as the Backbone network has been considered. The IRL block has been designed for minimal impact on computational speed. By examining this component, specifically for the CUHK03 dataset, there was a 5% increase in mAP and a 3% increase in @Rank1. For the Market-1501 dataset, the improvement is partially evident. Comparisons with more complex architectures have shown a significant increase in computational speed in these methods.ConclusionReducing computational costs while increasing relative recognition accuracy are interdependent objectives. Depending on the specific context and priorities, one might emphasize one over the other when selecting an appropriate method. The changes applied in this research can lead to more optimal results in method selection, striking a balance between computational efficiency and recognition accuracy.Keywords: Person Re-Identification, Deep Learning, Convolutional Neural Network, Image Detection
-
An optimization algorithm based on training and learning is formed based on the process of training and learning in a class. A deep neural network is one of the types of feedforward neural networks whose connection pattern among its neurons is inspired by the visual cortex of animals' brain. The present study considers decreasing prediction error for the types of time series and the uncertainty in estimation parameters, improving the structure of the deep neural network and increasing response speed in the proposed neural network method; besides, the competitive performance and the collaboration among the neurons of deep neural network are also increased. Selected data is related to Qeshm weather (suitable weather conditions to study our purpose) prediction during 2016 onwards. In this study, for the purpose of analyzing the prediction issue of power consumption of domestic expenses in the indefinite and severe fluctuation mode, we decided to combine two methods of Long Short-Term Memory and Convolutional Neural Network. For the training of the deep network, the BP algorithm is used.Keywords: Optimization algorithm, time series, Estimation, Prediction, Convolutional Neural Network, Long Short-Term Memory
- نتایج بر اساس تاریخ انتشار مرتب شدهاند.
- کلیدواژه مورد نظر شما تنها در فیلد کلیدواژگان مقالات جستجو شدهاست. به منظور حذف نتایج غیر مرتبط، جستجو تنها در مقالات مجلاتی انجام شده که با مجله ماخذ هم موضوع هستند.
- در صورتی که میخواهید جستجو را در همه موضوعات و با شرایط دیگر تکرار کنید به صفحه جستجوی پیشرفته مجلات مراجعه کنید.