Deep learning-based classification and detection algorithms have emerged as a powerful tool for vehicle in intelligent transportation systems. The limitations of the number high-quality labeled training samples makes single methods incapable accomplishing acceptable accuracy road detection. This paper presents vehicles on publicly available datasets by utilizing YOLO-v5 architecture. paper’s findings utilize concept transfer learning through fine tuning weights pre-trained To employ...

The diversion of a driver’s attention from driving can be catastrophic. Given that conventional button- and touch-based interfaces may distract the driver, developing novel distraction-free for various devices present in cars has becomes necessary. Hand gesture recognition provide an alternative interface inside cars. are targeted application area, we determined optimal location radar sensor, so signal reflected hand during gesturing is unaffected by interference motion body or other motions...

Remote sensing image classification has great advantages in the areas of environmental monitoring, urban planning, disaster management and many others. Unmanned Aerial Vehicles (UAVs) have revolutionized remote by providing high-resolution imagery. In this context, effective is crucial for extracting meaningful information from UAV-captured images. This study presents a comparison different deep learning-based approach supervised UAV We experimented on four CNN models like VGG 16, Alex net,...

Abstract Aim The study was performed to evaluate factors influencing postoperative adverse events after Hartmann’s reversal (HR). Method This a retrospective of unselected patients who underwent HR the procedure (HP) for left colonic perforation with peritonitis at single institution. Data were retrieved from an Institutional Review Board‐approved database. end‐point events, which included mortality, complications, reoperations and 30‐day readmission. Lag time defined as HP HR. results are...

Nowadays many camera-based advanced driver assistance systems (ADAS) have been introduced to assist the drivers and ensure their safety under various driving conditions. One of problems faced by is faded scene visibility lower contrast while in foggy In this paper, we present a novel approach provide solution problem employing deep neural networks. We assume that fog an image can be mathematically modeled unknown complex function utilize network approximate corresponding mathematical model...

A compression technique for still digital images is proposed with deep neural networks (DNNs) employing rectified linear units (ReLUs). We tend to exploit the DNNs capabilities find a reasonable estimate of underlying compression/decompression relationships. aim DNN image purpose that has better generalization property and reduced training time support real operation. The use ReLUs which map more plausibly biological neurons, makes our significantly faster, shortens encoding/decoding time,...

Scale invariance and high miss detection rates for small objects are some of the challenging issues object often lead to inaccurate results. This research aims provide an accurate model crowd counting by focusing on human head from natural scenes acquired publicly available datasets Casablanca, Hollywood-Heads Scut-head. In this study, we tuned a yolov5, deep convolutional neural network (CNN) based architecture, then evaluated using mean average precision (mAP) score, precision, recall. The...

In this paper, a novel rotation and scale invariant approach for texture classification based on Gabor filters has been proposed. These are designed to capture the visual content of images their impulse responses which sensitive scaling in images. The filter rearranged according exhibiting response having largest amplitude, followed by calculation patterns after binarizing particular threshold. This threshold is obtained as average energy at pixel. binary converted decimal numbers,...

Infrastructure-less (sometimes known as ad-hoc) networking paradigm is very appealing and potentially shaping its future into almost all emerging networks (i.e., IoT, wireless sensor networks, vehicular ad-hoc emergency, tactical radio etc.). However, when conventional protocols are used, such often perform poorly, mainly because of interference within the network limited throughput. Recent advancements in communications have enabled full-duplex (FD) operation by suppressing...

Deep neural networks (DNNs) are increasingly being researched and employed as a solution to various image video processing tasks. In this paper we address the problem of digital compression using DNNs. We use two different DNN architectures for i.e. one employing logistic sigmoid neurons other engaging hyperbolic tangent neurons. Experiments show that network out performs with Results indicate not only improve PSNR reconstructed images by significant 2∼5dB on average but they also converge...

Determining risk factors for posterior capsule opacification will allow further interventions to reduce the of development and thus additional procedures.The purpose this study was investigate associated with clinically significant requiring yttrium aluminum garnet (YAG) capsulotomy.Medical records patients (≥18 years) who underwent cataract surgery between January 1, 2011, March 31, 2014, at Kresge Eye Institute were reviewed. Three hundred eyes YAG capsulotomy up 3 years after included in...

Introduction: Convolutional neural networks (CNNs) have maintained their dominance in deep learning methods for human action recognition (HAR) and other computer vision tasks. However, the need a large amount of training data always restricts performance CNNs. Method: This paper is inspired by two-stream network, where CNN deployed to train network using spatial temporal aspects an activity, thus exploiting strengths both achieve better accuracy. Contributions: Our contribution twofold:...

Coronary artery disease (CAD) is one of the most common causes sudden cardiac arrest, accounting for a large percentage global mortality. A timely diagnosis and detection may save person's life. The research suggests methodological framework non-invasive risk stratification based on information only possible after invasive coronary angiography. Novel clinical, chemical, molecular biomarkers were used as input features from an especially collected dataset. Following thorough evaluative search...

This paper presents a novel framework for fraud detection in healthcare systems which self-learns from the historical medical data. Historical records are required training and testing of machine learning models. The main problem being faced by both private government health supported schemes is rapid rise amount claims beneficiaries mostly based on fraudulent billing. Detection transactions strenuous task due to intricate relationships among dynamic elements including doctors, patients,...

The paper presents a novel methodology based on machine learning to optimize medical benefits in healthcare settings, i.e., corporate, private, public or statutory. optimization is applied design insurance packages the employee record. Moreover, with advancement industry, it rapidly adapting mathematical and models enhance services like funds prediction, customer management get better revenue from their businesses. However, conventional computing premium methods are time-consuming,...

Educational Data Mining (EDM) has become one of the most important fields now a day because with development technology, student's problems are also increasing. In order to tackle these and help students, educational data mining come into existence. this research paper, Systematic Literature Review (SLR) been carried out get 20 studies (2012-2019) in field EDM. From studies, 11 highly advanced machine learning models have obtained we implemented them on 2 public student databases predict...

A new utility function‐based framework is proposed for network selection that utilises exponential functions and monotonically related to a function of the average additive multiplicative functions. This means an intermediate solution has been identified not only includes inter‐dependencies performance criteria but also reduces elimination. Numerical results compare existing it can be seen good choice selection.

Abstract Imaging systems with different imaging sensors are widely applied to surveillance field, military and medicine field. Particularly, infrared can acquire thermal radiations emitted by objects but lack textural details, visible capture abundant information suffer from loss of scene under poor weather conditions. The fusion images synthesize a new image complementary the source images. In this paper, we present deep learning method encoder–decoder architecture for fusion. Firstly,...