جستجوی مقالات مرتبط با کلیدواژه "convolutional neural network" در نشریات گروه "برق"

Re-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. Since 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. Given 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. Reducing 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.

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.

Fingerprint verification has emerged as a cornerstone of personal identity authentication. This research introduces a deep learning-based framework for enhancing the accuracy of this critical process. By integrating a pre-trained Inception model with a custom-designed architecture, we propose a model that effectively extracts discriminative features from fingerprint images. To this end, the input fingerprint image is aligned to a base fingerprint through minutiae vector comparison. The aligned input fingerprint is then subtracted from the base fingerprint to generate a residual image. This residual image, along with the aligned input fingerprint and the base fingerprint, constitutes the three input channels for a pre-trained Inception model. Our main contribution lies in the alignment of fingerprint minutiae, followed by the construction of a color fingerprint representation. Moreover, we collected a dataset, including 200 fingerprint images corresponding to 20 persons, for fingerprint verification. The proposed method is evaluated on two distinct datasets, demonstrating its superiority over existing state-of-the-art techniques. With a verification accuracy of 99.40% on the public Hong Kong Dataset, our approach establishes a new benchmark in fingerprint verification. This research holds the potential for applications in various domains, including law enforcement, border control, and secure access systems.

Fungal infections, capable of establishing in various tissues and organs, are responsible for many human diseases that can lead to serious complications. The initial step in diagnosing fungal infections typically involves the examination of microscopic images. Direct microscopic examination using potassium hydroxide is commonly employed as a screening method for diagnosing superficial fungal infections. Although this type of examination is quicker than other diagnostic methods, the evaluation of a complete sample can be time-consuming. Moreover, the diagnostic accuracy of these methods may vary depending on the skill of the practitioner and does not guarantee full reliability. This paper introduces a novel approach for diagnosing fungal infections using a modified VGG19 deep learning architecture. The method incorporates two significant changes: replacing the Flatten layer with Global Average Pooling (GAP) to reduce feature count and model complexity, thereby enhancing the extraction of significant features from images. Additionally, a Dense layer with 1024 neurons is added post-GAP, enabling the model to better learn and integrate these features. The Defungi microscopic dataset was used for training and evaluating the model. The proposed method can identify fungal diseases with an accuracy of 97%, significantly outperforming the best existing method, which achieved an accuracy of 92.49%. This method not only significantly outperforms existing methods, but also, given its high accuracy, is valuable in the field of diagnosing fungal infections. This work demonstrates that the use of deep learning in diagnosing fungal diseases can lead to a substantial improvement in the quality of health services.

This paper provides a review of deep learning-based methods for fault diagnosis of electrical motors. Electrical motors are crucial components in various industrial applications, and their efficient operation is essential for maintaining productivity and minimizing downtime. Traditional fault diagnosis methods have limitations in accurately detecting and classifying motor faults. Deep learning, a subset of machine learning, has emerged as a promising approach for improving fault diagnosis accuracy. This review discusses various deep learning methods, such as convolutional neural networks, recurrent neural networks, autoencoders, transfer learning, and transformers that have been utilized for motor fault diagnosis. Additionally, it examines different datasets and features used in these methods, highlighting their advantages and limitations. The paper also discusses challenges and future research directions in this field, such as data augmentation, transfer learning, and interpretability of deep learning models. Based on the findings, it is concluded that deep learning-based technologies are replacing manual expert involvement as the new norms in this field. Besides, methods are getting more standard, and official benchmarks are being created. A summarized table is provided at the end of the paper and numerous methods have been reported.

The heating, ventilation, and air conditioning (HVAC) control system is responsible for the efficient building energy system. Indoor energy consumption patterns can be monitored and reduced intelligently. Occupancy information plays a vital role to save a reasonable amount of energy. Traditional energy monitoring and control systems can be improved with the installation of the occupancy monitoring system which will consist of a network of sensors and cameras. In this research work, we propose a new and revolutionary convolutional neural network (CNN) based on real-time camera occupancy detection and recognition techniques across different sorts of sensors that provide realistic low-cost energy-saving solutions with robust graphical processing units (GPUs). This occupancy information will decide the energy behaviour inside buildings. Decision-making tools can be used to select the appropriate occupancy detection and recognition alternative for indoor environment and energy monitoring and management. In this research work, we introduce and develop the "Fermatean fuzzy prioritized weighted average and geometric operator". These aggregation operators (AOs) are a modern approach to modelling complexities in decision-making. In the end, we give an algorithm for an intelligent decision support system (IDSS) using proposed AOs to compare our CNN based method with other existing sensors techniques.

Breast cancer is a disease of abnormal cell proliferation in the breast tissue organs. One method for diagnosing and screening breast cancer is mammography. However, the results of this mammography image have limitations because it has low contrast and high noise and contrast as non-coherence. This research segmented breast cancer images derived from Ultrasonography (USG) photo using a Convolutional Neural Network (CNN) using the U-Net architecture. Testing on the CNN model with the U-Net architecture results the highest Mean Intersection over Union (Mean IoU) value in the data scenario with a ratio of 70:30, 100 epochs, and a learning rate of 5x10-5, which is 77%, while the lowest Mean IoU in the data scenario with a ratio 90:10, 50 epochs, and a learning rate of 1x10-4 learning rate, which is 64.4%.

The coronavirus disease or COVID-19, as a global disease, is an unprecedented health care crisis due to increasing mortality and its high rate of infection. Patients usually show significant complications in the respiratory system. This disease is caused by SARS-CoV-2. Decreasing the time of diagnosis is essential for reducing deaths and low spreading of the virus. Also, using the optimal tool in the pediatric setting and Intensive care unit (ICU) is required. Therefore, using lung ultrasound is recommended. It does not have any radiation and it has a lower cost. However, it makes noisy and low-quality data. In this paper, we propose a novel approach called Uniform Local Binary Pattern on Five intersecting Planes and convolutional neural Network (ULBPFP-Net) that overcomes the said limitation. We extract worthwhile features from five planes for feeding a network. Our experiments confirm the success of the ULBPFP-Net in COVID-19 diagnosis compared to the previous approaches.

Human activity recognition (HAR) has the potential to significantly impact applications such as health monitoring, context-aware systems, transportation, robotics, and smart cities. Because of the prevalence of wireless devices, the Wi-Fi-based approach has attracted a lot of attention among other existing methods such as sensor-based and vision-based HAR. Wi-Fi devices can be used to distinguish between daily activities such as "walking," "running," and "sleeping," which affect Wi-Fi signal propagation. This paper proposes a Deep Learning method for HAR tasks that makes use of channel state information (CSI). We convert the CSI data to RGB images and classify the activity recognition using a 2D-Convolutional Neural Network (CNN). We evaluate the performance of the proposed method on two publicly available datasets for CSI data. Our experiments show that converting data into RGB images improves performance and accuracy compared to our previous method by at least 5%.

Handwritten character recognition has occupied a substantial area due to its applications in several fields and is used widely in the modern world. Handwritten Arabic recognition is a major challenge because of the high similarity in its characters and its various writing styles. Deep learning algorithms have recently shown high performance in this area. The problem is that a deep learning algorithm requires large datasets for training. To overcome this problem, an efficient architecture is presented in this study, which comprises Hidden Markovian Model for character modeling, Convolutional Neural Network for feature extraction, and an intelligent network for recognition. The proposed network is modeled based on the dentate gyrus of the hippocampus of the brain. This part of the brain is responsible for identifying highly overlapping samples. The handwritten Arabic alphabet is characterized by this high overlap. Modeling the functionality of the dentate gyrus can improve the accuracy of the handwritten Arabic characters. Experiments are done using IFN/ENIT, CMATERdb3.3.1 and, MADBase datasets. The proposed approach outperformed recently published works using the same dataset. Although in all the experiments, a character error rate (CER) of less than 1.63 was obtained, the CMATERdb3.3.1 dataset resulted in a CER of 0.35. Compared with the convolutional neural network, the proposed network showed superiority in recognizing patterns with high noise.

Retinal blood vessels include arteries and veins and are usually next to each other. Blood vessels are used to classify the severity of the disease and are also used for guidance during surgery, as retinopathy is one of the dangerous diseases.Diabetic retinopathy can cause the formation of new vessels (neoangiogenesis). This condition causes low vision and even blindness. Therefore, a reliable method for diagnosing and classifying the vessel is needed in order to avoid these complications. Retinopathy is one of the hidden diseases that is usually not known. prevent the next possibility.There are several methods for diagnosis, the most common of which is the use of traditional methods based on manual feature extraction, which requires a lot of feature geometry and expertise, and is usually dependent on data.From this method, neural convolution is a reliable, efficient and reliable method for extracting features without manual intervention, which requires a lot of expertise, which also reduces the dependence on data.In this article, using convolutional neural network, diabetic retinopathy has been diagnosed with accuracy and sensitivity of 98.8% and 97.5%, respectively.The obtained results indicate that the proposed method is suitable for locating blood vessels automatically.

Deep learning methods use neural networks that try to discover patterns within the image without human intervention and to learn that. One of the most popular algorithms in this field is the convolutional neural network algorithm. This algorithm uses several layers to receive the input image and process it so that the class label can be found. These layers are mostly based on Neural Networks. This research aims to provide a model of neural-fuzzy, based on convolutional neural network algorithm. In this research, we use the positive advantages of deep learning methods and fuzzy inference systems and present a new model of their application to Classify authorized and unauthorized Persons. For this purpose, we designed new neural-fuzzy layers to pass the image through them and finally classify each image. The results of the implementation of the above model show the efficiency and success of this system.

Super resolution algorithms attempt to reconstruct high resolution images from low resolution images and it can be considered as a preprocessing step for object recognition and image classification. Various algorithms have been introduced for single-image super resolution, but these algorithms often face important challenges such as poorly matching the reconstructed image with the original image. This article introduces a preprocessing operation to improve the performance of the super resolution process. In the proposed method, the low-resolution images are enhanced before entering to the resolution change module. Calculating the brightness of the pixels in the image channels, creating the luminance map and removing atmospheric light, applying the transmittance map by using the luminance coefficients, and recovering the natural image in all three-color channels are the above preprocessing steps. The proposed method succeeded in increasing the PSNR parameter by 4.35%, 10.62%, and 8.31%, as well as 0.23%, 3.10%, and 7.91% of the SSIM parameter for Set5, Set14, and BSD100 benchmark datasets compared to its closest state-of-the-art methods.