Search results for: computational vision

Authors: Mahdi Hamzehei, Homan Alimoradzadeh, Mahdi Shahriyari

Abstract:

In this paper, vibration of the industrial centrifugal pumps studied by measuring analysis and computational simulation. Effects of different parameters on pump vibration were investigated. Also, simulation of cavitation in the centrifugal pump was down. First, via CF-TURBO software, the pump impeller and the fluid passing through the pump is modelled and finally, the phenomenon of cavitation in the impeller has been modelled by Ansys software. Also, the effects of changes in the amount of NPSH and bubbles generation in the pump impeller were investigated. By simulation of piping with pipe flow software, effect of fluid velocity and pressure on hydraulics and vibration were studied computationally by applying Computational Fluid Dynamic (CFD) techniques, fluent software and experimentally. Furthermore, this comparison showed that the model can predict hydraulics and vibration behaviour.

Keywords: cavitation, vibration, centrifugal pumps, performance curves, NPSH

Procedia PDF Downloads 543

Authors: Xingdu Wang, Charlie Gere, Emma Rose, Yuxuan Zhao

Abstract:

The study introduces a fresh perspective on the digital realm through an examination of the Chinese concept of Xiang, elucidating how it can build an understanding of pixels and textures on screens as digital trigrams. This concept attempts to offer an outlook on the intersection of digital technology and the natural world, thereby contributing to discussions about the harmonious relationship between humans and technology. The study looks for the ancient Chinese theory of Xiang as a key to establishing the theories and practices to respond to the problem of Contemporary Chinese technics. Xiang is a Chinese method of understanding the essentials of things through appearances, which differs from the method of science in the Westen. Xiang, the basement of Chinese visual art, is rooted in ancient Chinese philosophy and connected to the eight trigrams. The discussion of Xiang connects art, philosophy, and technology. This paper connects the meaning of Xiang with the 'truth appearing' philosophically through the analysis of the concepts of phenomenon and noumenon and the unique Chinese way of observing. Hereafter, the historical interconnection between ancient painting and writing in China emphasizes their relationship between technical craftsmanship and artistic expression. In digital, the paper blurs the traditional boundaries between images and text on digital screens in theory. Lastly, this study identified an ensemble concept relating to pixels and textures in computer vision, drawing inspiration from AI image recognition in Chinese paintings. In art practice, by presenting a fluid visual experience in the form of pixels, which mimics the flow of lines in traditional calligraphy and painting, it is hoped that the viewer will be brought back to the process of the truth appearing as defined by the 'Xiang’.

Keywords: Chinese cosmotechnics, computer vision, contemporary Neo-Confucianism, texture and pixel, Xiang

Procedia PDF Downloads 64

Authors: Tae-min Byun, Chang-soo Chon, Dong-hyun Seo, Han-sung Kim, Bum-mo Ahn, Hui-suk Yun, Cheolwoong Ko

Abstract:

In this study, a 3-point bending computational analysis of normal and osteoporosis mouse models was performed based on the Micro-CT image information of the femurs. The finite element analysis (FEA) found 1.68 N (normal group) and 1.39 N (osteoporosis group) in the average maximum force, and 4.32 N/mm (normal group) and 3.56 N/mm (osteoporosis group) in the average stiffness. In the comparison of the 3-point bending test results, the maximum force and the stiffness were different about 9.4 times in the normal group and about 11.2 times in the osteoporosis group. The difference between the analysis and the test was greatly significant and this result demonstrated improvement points of the material properties applied to the computational analysis of this study. For the next study, the material properties of the mouse femur will be supplemented through additional computational analysis and test.

Keywords: 3-point bending test, mouse, osteoporosis, FEA

Procedia PDF Downloads 351

Authors: Michel Wakim, Rodrigo Rivera Tinoco

Abstract:

Supersonic ejector is an economical device that use high pressure vapor to compress a low pressure vapor without any rotating parts or external power sources. Entrainment ratio is a major characteristic of the ejector performance, so the ejector performance is highly dependent on its geometry. The aim of this paper is to design ejector geometry, based on pre-specified operating conditions, and to study the flow behavior inside the ejector by using computational fluid dynamics ‘CFD’ by using ‘ANSYS FLUENT 15.0’ software. In the first section; 1-D mathematical model is carried out to predict the ejector geometry. The second part describes the flow behavior inside the designed model. CFD is the most reliable tool to reveal the mixing process at different parts of the supersonic turbulent flow and to study the effect of the geometry on the effective ejector area. Finally, the results show the effect of the geometry on the entrainment ratio.

Keywords: computational fluids dynamics, ejector, entrainment ratio, geometry optimization, performance

Procedia PDF Downloads 274

Authors: Zair Mohammed El Amine, Chemouri Hafida, Derbal Habak Hassina

Abstract:

Organic photovoltaic cells are electronic devices that convert sunlight into electricity. To this end, the number of studies on organic photovoltaic cells (OVCs) is growing, and this trend is expected to continue. Computational studies are still needed to verify and prove the capability of CVOs, specifically the nanometer molecule PCBM, based on successful experimental results. In this paper, we present a theoretical and computational investigation of PCBM and PC71BM derivatives using the DFT method. On this basis, we employ independent and time-dependent density theories. HOMO, LUMO and GAPH-L energies, ionization potentials and electronic affinity are determined and found to be in agreement with experiments. Using DFT theory based on B3LYP and M062X methods with bases 6-31G (d,p) and 6-311G (d), calculations show that the most efficient acceptors are presented in the group of PC71BM derivatives and are in substantial agreement with experiments. The geometries of the structures are optimized by Gaussian 09.

Keywords: PCBM, P3HT, organic cell solar, DFT, TD-DFT

Procedia PDF Downloads 86

Authors: Promise Ekpo Osaine, Daniel Melesse

Abstract:

Image captioning is an active area of research in the multi-modal artificial intelligence (AI) community as it connects vision and language understanding, especially in settings where it is required that a model understands the content shown in an image and generates semantically and grammatically correct descriptions. In this project, we followed a standard approach to a deep learning-based image captioning model, injecting architecture for the encoder-decoder setup, where the encoder extracts image features, and the decoder generates a sequence of words that represents the image content. As such, we investigated image encoders, which are ResNet101, InceptionResNetV2, EfficientNetB7, EfficientNetV2M, and CLIP. As a caption generation structure, we explored long short-term memory (LSTM). The CLIP-LSTM model demonstrated superior performance compared to the encoder-decoder models, achieving a BLEU-1 score of 0.904 and a BLEU-4 score of 0.640. Additionally, among the CNN-LSTM models, EfficientNetV2M-LSTM exhibited the highest performance with a BLEU-1 score of 0.896 and a BLEU-4 score of 0.586 while using a single-layer LSTM.

Keywords: multi-modal AI systems, image captioning, encoder, decoder, BLUE score

Procedia PDF Downloads 77

Authors: Aline Giardin, Maria Rosa Chitolina, Maria Catarina Zanini

Abstract:

In this paper, we analyze the conceptions of high school students about mental health issues, and discuss the creation of mental basic health programs in schools. We base our findings in a quantitative survey carried out by us with 156 high school students of CTISM (Colégio Técnico Industrial de Santa Maria) school, located in Santa Maria city, Brazil. We have found that: (a) 28 students relate the subject ‘mental health’ with psychiatric hospitals and lunatic asylums; (b) 28 students have relatives affected by mental diseases; (c) 76 students believe that mental patients, if treated, can live a healthy life; (d) depression, schizophrenia and bipolar disorder are the most cited diseases; (e) 84 students have contact with mental patients, but know nothing about the disease; (f) 123 students have never been instructed about mental diseases while in the school; and (g) 135 students think that a mental health program would be important in the school. We argue that these numbers reflect a vision of mental health that can be related to the reductionist education still present in schools and to the lack of integration between health professionals, sciences teachers, and students. Furthermore, this vision can also be related to a stigmatization process, which interferes with the interactions and with the representations regarding mental disorders and mental patients in society.

Keywords: mental health, schools, mental illness, conception

Procedia PDF Downloads 469

Authors: Jamshaid Ul Rahman, Faiza Makhdoom, Dianchen Lu

Abstract:

Activation functions play a decisive role in determining the capacity of Deep Neural Networks (DNNs) as they enable neural networks to capture inherent nonlinearities present in data fed to them. The prior research on activation functions primarily focused on the utility of monotonic or non-oscillatory functions, until Growing Cosine Unit (GCU) broke the taboo for a number of applications. In this paper, a Convolutional Neural Network (CNN) model named as ASU-CNN is proposed which utilizes recently designed activation function ASU across its layers. The effect of this non-monotonic and oscillatory function is inspected through feature map visualizations from different convolutional layers. The optimization of proposed network is offered by Adam with a fine-tuned adjustment of learning rate. The network achieved promising results on both training and testing data for the classification of CIFAR-10. The experimental results affirm the computational feasibility and efficacy of the proposed model for performing tasks related to the field of computer vision.

Keywords: amplifying sine unit, activation function, convolutional neural networks, oscillatory activation, image classification, CIFAR-10

Procedia PDF Downloads 110

Authors: Y. Bablu Singh, B. S. Purkayashtha, Chungkham Yashawanta Singh

Abstract:

Morphological analysis forms the basic foundation in NLP applications including syntax parsing Machine Translation (MT), Information Retrieval (IR) and automatic indexing in all languages. It is the field of the linguistics; it can provide valuable information for computer based linguistics task such as lemmatization and studies of internal structure of the words. Computational Morphology is the application of morphological rules in the field of computational linguistics, and it is the emerging area in AI, which studies the structure of words, which are formed by combining smaller units of linguistics information, called morphemes: the building blocks of words. Morphological analysis provides about semantic and syntactic role in a sentence. It analyzes the Manipuri word forms and produces several grammatical information associated with the words. The Morphological Analyzer for Manipuri has been tested on 3500 Manipuri words in Shakti Standard format (SSF) using Meitei Mayek as source; thereby an accuracy of 80% has been obtained on a manual check.

Keywords: morphological analysis, machine translation, computational morphology, information retrieval, SSF

Procedia PDF Downloads 326

Authors: Samah Laalej, Abdelfattah Bouatem

Abstract:

In the field of wind turbine blades, it is complicated to evaluate the aerodynamic performances through experimental measurements as it requires a lot of computing time and resources. Therefore, in this paper, a hybrid BEM-CFD numerical technique is developed to predict power and aerodynamic forces acting on the blades. Computational fluid dynamics (CFD) simulation was conducted to calculate the drag and lift forces through Ansys software using the K-w model. Then an enhanced BEM code was created to predict the power outputs generated by the wind turbine using the aerodynamic properties extracted from the CFD approach. The numerical approach was compared and validated with experimental data. The power curves calculated from this hybrid method were in good agreement with experimental measurements for all velocity ranges.

Keywords: blade element momentum, aerodynamic forces, wind turbine blades, computational fluid dynamics approach

Procedia PDF Downloads 64

Authors: Shunyi Yang, Jingjing Yan, Siyu Dong, Xiangguo Cui

Abstract:

Current architectural façade design practices incorporate various daylighting and solar radiation analysis methods. These emphasize the impact of geometry on façade design. There is scope to extend this knowledge into methods that address material translucency, porosity, and form. Such approaches can also achieve these conditions through adaptive robotic manufacturing approaches that exploit material dynamics within the design, and alleviate fabrication waste from molds, ultimately accelerating the autonomous manufacturing system. Besides analyzing the environmental solar radiant in building facade design, there is also a vacancy research area of how lighting effects can be precisely controlled by engaging the instant real-time data-driven robot control and manipulating the material properties. Ceramics carries a wide range of transmittance and deformation potentials for robotics control with the research of its material property. This paper presents one semi-autonomous system that engages with real-time data-driven robotics control, hardware kit design, environmental building studies, human interaction, and exploratory research and experiments. Our objectives are to investigate the relationship between different clay bodies or ceramics’ physio-material properties and their transmittance; to explore the feedback system of instant lighting data in robotic fabrication to achieve precise lighting effect; to design the sufficient end effector and robot behaviors for different stages of deformation. We experiment with architectural clay, as the material of the façade that is potentially translucent at a certain stage can respond to light. Studying the relationship between form, material properties, and porosity can help create different interior and exterior light effects and provide façade solutions for specific architectural functions. The key idea is to maximize the utilization of in-progress robotics fabrication and ceramics materiality to create a highly integrated autonomous system for lighting facade design and manufacture.

Keywords: light transmittance, data-driven fabrication, computational design, computer vision, gamification for manufacturing

Procedia PDF Downloads 123

Authors: Momoh Omeiza Sheidu

Abstract:

Finite element method as a method of providing solutions to problems in computational bio mechanics provides a framework for modeling the function of tissues that integrates structurally from cell to organ system and functionally across the physiological processes that affect tissue mechanics or are regulated by mechanical forces. In this paper, we present an integrative finite element strategy for solution to problems in tissue bio mechanics as a case study.

Keywords: finite element, biomechanics, modeling, computational biomechanics

Procedia PDF Downloads 503

Authors: Babar Khan, Wang Zhijie

Abstract:

Machine Vision has been playing a significant role in Industrial Automation, to imitate the wide variety of human functions, providing improved safety, reduced labour cost, the elimination of human error and/or subjective judgments, and the creation of timely statistical product data. Despite the intensive research, there have not been any attempts to classify fabric prints based on printed texture and colour, most of the researches so far encompasses only black and white or grey scale images. We proposed a biologically inspired processing architecture to classify fabrics w.r.t. the fabric print texture and colour. We created a texture descriptor based on the HMAX model for machine vision, and incorporated colour descriptor based on opponent colour channels simulating the single opponent and double opponent neuronal function of the brain. We found that our algorithm not only outperformed the original HMAX algorithm on classification of fabric print texture and colour, but we also achieved a recognition accuracy of 85-100% on different colour and different texture fabric.

Keywords: automatic classification, texture descriptor, colour descriptor, opponent colour channel

Procedia PDF Downloads 484

Authors: Cagri Memis, Muzaffer Kapanoglu

Abstract:

The school bus routing problem (SBRP) is a variant of the Vehicle Routing Problem (VRP) classified as a location-allocation-routing problem. In this study, the SBRP is decomposed into two sub-problems: (1) bus route generation and (2) bus stop selection to solve large instances of the SBRP in reasonable computational times. To solve the first sub-problem, we propose a genetic algorithm to generate bus routes. Once the routes have been fixed, a sub-problem remains of allocating students to stops considering the capacity of the buses and the walkability constraints of the students. While the exact method solves small-scale problems, treating large-scale problems with the exact method becomes complex due to computational problems, a deficiency that the genetic algorithm can overcome. Results obtained from the proposed approach on 150 instances up to 250 stops show that the matheuristic algorithm provides better solutions in reasonable computational times with respect to benchmark algorithms.

Keywords: genetic algorithm, matheuristic, school bus routing problem, vehicle routing problem

Procedia PDF Downloads 71

Authors: Yifan Lu, Ying Cui, Ying Zhu, Edward S. Lu, Rebecca Zeng, Rohan Bajaj, Raviv Katz, Rongrong Le, Jay C. Wang, John B. Miller

Abstract:

Purpose: Retinal artery occlusion (RAO) is an ophthalmic emergency that can lead to poor visual outcome and is associated with an increased risk of cerebral stroke and cardiovascular events. Fluorescein angiography (FA) is the traditional diagnostic tool for RAO; however, wide-field swept-source optical coherence tomography angiography (WF SS-OCTA), as a nascent imaging technology, is able to provide quick and non-invasive angiographic information with a wide field of view. In this study, we looked for associations between OCT-A vascular metrics and visual acuity in patients with prior diagnosis of RAO. Methods: Patients with diagnoses of central retinal artery occlusion (CRAO) or branched retinal artery occlusion (BRAO) were included. A 6mm x 6mm Angio and a 15mm x 15mm AngioPlex Montage OCT-A image were obtained for both eyes in each patient using the Zeiss Plex Elite 9000 WF SS-OCTA device. Each 6mm x 6mm image was divided into nine Early Treatment Diabetic Retinopathy Study (ETDRS) subfields. The average measurement of the central foveal subfield, inner ring, and outer ring was calculated for each parameter. Non-perfusion area (NPA) was manually measured using 15mm x 15mm Montage images. A linear regression model was utilized to identify a correlation between the imaging metrics and visual acuity. A P-value less than 0.05 was considered to be statistically significant. Results: Twenty-five subjects were included in the study. For RAO eyes, there was a statistically significant negative correlation between vision and retinal thickness as well as superficial capillary plexus vessel density (SCP VD). A negative correlation was found between vision and deep capillary plexus vessel density (DCP VD) without statistical significance. There was a positive correlation between vision and choroidal thickness as well as choroidal volume without statistical significance. No statistically significant correlation was found between vision and the above metrics in contralateral eyes. For NPA measurements, no significant correlation was found between vision and NPA. Conclusions: This is the first study to our best knowledge to investigate the utility of WF SS-OCTA in RAO and to demonstrate correlations between various retinal vascular imaging metrics and visual outcomes. Further investigations should explore the associations between these imaging findings and cardiovascular risk as RAO patients are at elevated risk for symptomatic stroke. The results of this study provide a basis to understand the structural changes involved in visual outcomes in RAO. Furthermore, they may help guide management of RAO and prevention of cerebral stroke and cardiovascular accidents in patients with RAO.

Keywords: OCTA, swept-source OCT, retinal artery occlusion, Zeiss Plex Elite

Procedia PDF Downloads 139

Authors: Parampal Kaur, Deepak Aggarwal

Abstract:

In the computational Grid, fault tolerance is an imperative issue to be considered during job scheduling. Due to the widespread use of resources, systems are highly prone to errors and failures. Hence, fault tolerance plays a key role in the grid to avoid the problem of unreliability. Scheduling the task to the appropriate resource is a vital requirement in computational Grid. The fittest resource scheduling algorithm searches for the appropriate resource based on the job requirements, in contrary to the general scheduling algorithms where jobs are scheduled to the resources with best performance factor. The proposed method is to improve the fault tolerance of the fittest resource scheduling algorithm by scheduling the job in coordination with job replication when the resource has low reliability. Based on the reliability index of the resource, the resource is identified as critical. The tasks are scheduled based on the criticality of the resources. Results show that the execution time of the tasks is comparatively reduced with the proposed algorithm using real-time approach rather than a simulator.

Keywords: computational grid, fault tolerance, task replication, job scheduling

Procedia PDF Downloads 436

Authors: A. Mohammadi, M. Moghimi, S. Mohammadi

Abstract:

Engineers need computational methods which could provide solutions less sensitive to the environmental effects, so the techniques should be used which take the uncertainty to account to control and minimize the risk associated with design and operation. In order to consider uncertainty in engineering problem, the optimization problem should be solved for a suitable range of the each uncertain input variable instead of just one estimated point. Using deterministic optimization problem, a large computational burden is required to consider every possible and probable combination of uncertain input variables. Several methods have been reported in the literature to deal with problems under uncertainty. In this paper, different methods presented and analyzed.

Keywords: uncertainty, Monte Carlo simulated, stochastic programming, scenario method

Procedia PDF Downloads 414

Authors: Taweechai Nuntawisuttiwong, Natasha Dejdumrong

Abstract:

Newton-Lagrange Interpolations are widely used in numerical analysis. However, it requires a quadratic computational time for their constructions. In computer aided geometric design (CAGD), there are some polynomial curves: Wang-Ball, DP and Dejdumrong curves, which have linear time complexity algorithms. Thus, the computational time for Newton-Lagrange Interpolations can be reduced by applying the algorithms of Wang-Ball, DP and Dejdumrong curves. In order to use Wang-Ball, DP and Dejdumrong algorithms, first, it is necessary to convert Newton-Lagrange polynomials into Wang-Ball, DP or Dejdumrong polynomials. In this work, the algorithms for converting from both uniform and non-uniform Newton-Lagrange polynomials into Wang-Ball, DP and Dejdumrong polynomials are investigated. Thus, the computational time for representing Newton-Lagrange polynomials can be reduced into linear complexity. In addition, the other utilizations of using CAGD curves to modify the Newton-Lagrange curves can be taken.

Keywords: Lagrange interpolation, linear complexity, monomial matrix, Newton interpolation

Procedia PDF Downloads 234

Authors: Gebreegziabher Hailu

Abstract:

This paper presents the development of residual power series methods aimed at efficiently solving stiff differential equations, which pose significant challenges in numerical analysis due to their rapid changes in solution behavior. The RPSM is a numerical approach that generates polynomial-based approximate solutions without the need for linearization, discretization, or perturbation techniques, making it straightforward to implement and less prone to computational errors. We introduce an approach that utilizes power series expansions combined with residual minimization techniques to enhance convergence and stability. By analyzing the theoretical foundations of stiffness, we delve into the formulation of the residual power series method, detailing how it effectively captures the dynamics of stiff systems while maintaining computational efficiency. Numerical experiments demonstrate the method's superiority in terms of accuracy and computational cost when compared to traditional methods like implicit Runge-Kutta or multistep techniques. We also explore adaptive strategies within our framework to automatically adjust parameters based on the stiffness characteristics of the problem at hand. Ultimately, our findings contribute to the broader toolkit for tackling stiff differential equations, offering a robust alternative that promises to streamline computational workflows in various applied mathematics and engineering contexts.

Keywords: residual power series methods, stiff differential equoations, numerical approach, Runge Kutta methods

Procedia PDF Downloads 22

Authors: Simiao Ren, En Wei

Abstract:

Drowning is a significant safety issue worldwide, and a robust computer vision-based alert system can easily prevent such tragedies in swimming pools. However, due to domain shift caused by the visual gap (potentially due to lighting, indoor scene change, pool floor color etc.) between the training swimming pool and the test swimming pool, the robustness of such algorithms has been questionable. The annotation cost for labeling each new swimming pool is too expensive for mass adoption of such a technique. To address this issue, we propose a domain-aware data augmentation pipeline based on Gaussian Poisson Generative Adversarial Network (GP-GAN). Combined with YOLOv8, we demonstrate that such a domain adaptation technique can significantly improve the model performance (from 0.24 mAP to 0.82 mAP) on new test scenes. As the augmentation method only require background imagery from the new domain (no annotation needed), we believe this is a promising, practical route for preventing swimming pool drowning.

Keywords: computer vision, deep learning, YOLOv8, detection, swimming pool, drowning, domain adaptation, generative adversarial network, GAN, GP-GAN

Procedia PDF Downloads 100

Authors: Kashika P. H.

Abstract:

The development of computer vision systems has been greatly aided by the efficient and precise detection of images and videos. Although the ability to recognize and comprehend images is a strength of the human brain, employing technology to tackle this issue is exceedingly challenging. In the past few years, the use of Deep Learning algorithms to treat object detection has dramatically expanded. One of the key issues in the realm of image recognition is the recognition and detection of certain notable people from randomly acquired photographs. Face recognition uses a way to identify, assess, and compare faces for a variety of purposes, including user identification, user counting, and classification. With the aid of an accessible deep learning-based API, this article intends to recognize various faces of people and their facial descriptors more accurately. The purpose of this study is to locate suitable individuals and deliver accurate information about them by using the Amazon Rekognition system to identify a specific human from a vast image dataset. We have chosen the Amazon Rekognition system, which allows for more accurate face analysis, face comparison, and face search, to tackle this difficulty.

Keywords: Amazon rekognition, API, deep learning, computer vision, face detection, text detection

Procedia PDF Downloads 104

Authors: Hidouri Sami, Aguili Taoufik

Abstract:

We consider fast and accurate solutions of scattering problems by large perfectly conducting objects (PEC) formulated by an optimization of the Method of Auxiliary Sources (MAS). We present various techniques used to reduce the total computational cost of the scattering problem. The first technique is based on replacing the object by an array of finite number of small (PEC) object with the same shape. The second solution reduces the problem on considering only the half of the object.These two solutions are compared to results from the reference bibliography.

Keywords: method of auxiliary sources, scattering, large object, RCS, computational resources

Procedia PDF Downloads 241

Authors: Smitha Haridasan, Zelalem Demissie, Atri Dutta, Ajita Rattani

Abstract:

Aided by the wind, all it takes is one ember and a few minutes to create a wildfire. Wildfires are growing in frequency and size due to climate change. Wildfires and its consequences are one of the major environmental concerns. Every year, millions of hectares of forests are destroyed over the world, causing mass destruction and human casualties. Thus early detection of wildfire becomes a critical component to mitigate this threat. Many computer vision-based techniques have been proposed for the early detection of forest fire using video surveillance. Several computer vision-based methods have been proposed to predict and detect forest fires at various spectrums, namely, RGB, HSV, and YCbCr. The aim of this paper is to propose a multi-spectral deep learning model that combines information from different spectrums at intermediate layers for accurate fire detection. A heterogeneous dataset assembled from publicly available datasets is used for model training and evaluation in this study. The experimental results show that multi-spectral deep learning models could obtain an improvement of about 4.68 % over those based on a single spectrum for fire detection.

Keywords: deep learning, forest fire detection, multi-spectral learning, natural hazard detection

Procedia PDF Downloads 241

Authors: Massimo Miccoli, Luca Marangoni, Alberto Aniello Scaringi, Alessandro Marceddu, Alessandro Amicone

Abstract:

The susceptibility of deep neural networks (DNNs) to adversarial manipulations is a recognized challenge within the computer vision domain. Adversarial examples, crafted by adding subtle yet malicious alterations to benign images, exploit this vulnerability. Various defense strategies have been proposed to safeguard DNNs against such attacks, stemming from diverse research hypotheses. Building upon prior work, our approach involves the utilization of autoencoder models. Autoencoders, a type of neural network, are trained to learn representations of training data and reconstruct inputs from these representations, typically minimizing reconstruction errors like mean squared error (MSE). Our autoencoder was trained on a dataset of benign examples; learning features specific to them. Consequently, when presented with significantly perturbed adversarial examples, the autoencoder exhibited high reconstruction errors. The architecture of the autoencoder was tailored to the dimensions of the images under evaluation. We considered various image sizes, constructing models differently for 256x256 and 512x512 images. Moreover, the choice of the computer vision model is crucial, as most adversarial attacks are designed with specific AI structures in mind. To mitigate this, we proposed a method to replace image-specific dimensions with a structure independent of both dimensions and neural network models, thereby enhancing robustness. Our multi-modal autoencoder reconstructs the spectral representation of images across the red-green-blue (RGB) color channels. To validate our approach, we conducted experiments using diverse datasets and subjected them to adversarial attacks using models such as ResNet50 and ViT_L_16 from the torch vision library. The autoencoder extracted features used in a classification model, resulting in an MSE (RGB) of 0.014, a classification accuracy of 97.33%, and a precision of 99%.

Keywords: adversarial attacks, malicious images detector, binary classifier, multimodal transformer autoencoder

Procedia PDF Downloads 112

Authors: Omer Cahana, Maya Herman, Ofer Levi

Abstract:

Magnetic resonance imaging (MRI) is a long medical scan that stems from a long acquisition time. This length is mainly due to the traditional sampling theorem, which defines a lower boundary for sampling. However, it is still possible to accelerate the scan by using a different approach, such as compress sensing (CS) or parallel imaging (PI). These two complementary methods can be combined to achieve a faster scan with high-fidelity imaging. In order to achieve that, two properties have to exist: i) the signal must be sparse under a known transform domain, ii) the sampling method must be incoherent. In addition, a nonlinear reconstruction algorithm needs to be applied to recover the signal. While the rapid advance in the deep learning (DL) field, which has demonstrated tremendous successes in various computer vision task’s, the field of MRI reconstruction is still in an early stage. In this paper, we present an extension of the state-of-the-art model in MRI reconstruction -VarNet. We utilize VarNet by using dilated convolution in different scales, which extends the receptive field to capture more contextual information. Moreover, we simplified the sensitivity map estimation (SME), for it holds many unnecessary layers for this task. Those improvements have shown significant decreases in computation costs as well as higher accuracy.

Keywords: MRI, deep learning, variational network, computer vision, compress sensing

Procedia PDF Downloads 161

Authors: Nabila Jaman, K. E. Hoque, S. Sawall, M. Ferdows

Abstract:

Coronary artery disease (CAD) is one of the most lethal diseases of the cardiovascular diseases. Coronary arteries stenosis and bifurcation angles closely interact for myocardial infarction. We want to use computer-aided design model coupled with computational hemodynamics (CHD) simulation for detecting several types of coronary artery stenosis with different locations in an idealized model for identifying virtual fractional flow reserve (vFFR). The vFFR provides us the information about the severity of stenosis in the computational models. Another goal is that we want to imitate patient-specific computed tomography coronary artery angiography model for constructing our idealized models with different left anterior descending (LAD) and left circumflex (LCx) bifurcation angles. Further, we want to analyze whether the bifurcation angles has an impact on the creation of narrowness in coronary arteries or not. The numerical simulation provides the CHD parameters such as wall shear stress (WSS), velocity magnitude and pressure gradient (PGD) that allow us the information of stenosis condition in the computational domain.

Keywords: CAD, CHD, vFFR, bifurcation angles, coronary stenosis

Procedia PDF Downloads 157

Authors: Obulesh Avuku, Satwik Sunnam, Sri Charan Mohan Janthuka, Keerthi Yalamaddi

Abstract:

In the United States alone, there are over 1.5 million spine fractures per year, resulting in about 17,730 spinal cord injuries. The cervical spine is where fractures in the spine most frequently occur. The prevalence of spinal fractures in the elderly has increased, and in this population, fractures may be harder to see on imaging because of coexisting degenerative illness and osteoporosis. Nowadays, computed tomography (CT) is almost completely used instead of radiography for the imaging diagnosis of adult spine fractures (x-rays). To stop neurologic degeneration and paralysis following trauma, it is vital to trace any vertebral fractures at the earliest. Many approaches have been proposed for the classification of the cervical spine [2d models]. We are here in this paper trying to break the bounds and use the vision transformers, a State-Of-The-Art- Model in image classification, by making minimal changes possible to the architecture of ViT and making it 3D-enabled architecture and this is evaluated using a weighted multi-label logarithmic loss. We have taken this problem statement from a previously held Kaggle competition, i.e., RSNA 2022 Cervical Spine Fracture Detection.

Keywords: cervical spine, spinal fractures, osteoporosis, computed tomography, 2d-models, ViT, multi-label logarithmic loss, Kaggle, public score, private score

Procedia PDF Downloads 114

Authors: Hamza Rekab Djabri, Salah Daoud

Abstract:

The present paper deals with the computational of structural and electronic properties of ThC and ThN compounds using density functional theory within generalized-gradient (GGA) apraximation and local density approximation (LDA). We employ the full potential linear muffin-tin orbitals (FP-LMTO) as implemented in the Lmtart code. We have used to examine structure parameter in eight different structures such as in NaCl (B1), CsCl (B2), ZB (B3), NiAs (B8), PbO (B10), Wurtzite (B4) , HCP (A3) βSn (A5) structures . The equilibrium lattice parameter, bulk modulus, and its pressure derivative were presented for all calculated phases. The calculated ground state properties are in good agreement with available experimental and theoretical results.

Keywords: DFT, GGA, LDA, properties structurales, ThC, ThN

Procedia PDF Downloads 98

Authors: Sathyamangalam Rangasamy Guru Prasath

Abstract:

Due to the increasing homogeneity in product offerings, the attendant services provided are emerging as a key differentiator in the mind of the consumers. Services marketing are a sub field of marketing which covers the marketing of both goods and services. Service marketing differs from product marketing due to the face that services are intangible and typically require personal interaction with the customer. Relationships are a key factor when it comes to the marketing of services. The role of interpersonal relationships distinguishes service and product marketing in strategic vision and organizational considerations. This paper explores some of the trends in service marketing as they relate to strategic vision, operational and organizational changes, and marketing tactics. The presence of the customer in the service facility means that capacity management becomes an important driver of the firm’s profitability service marketing is a process from the organization’s point of view, but an experience from the customer’s perspective. The quality of the experience is a function of the careful design of customer service processes, adoption of standardized procedures, rigorous management of service quality, high standards of training and automation. Services marketing helps to ensure that these processes are designed from the customer’s perspective. Services marketing includes customer loyalty, managing relationships, complaint handling, improving service quality and productivity of service operations, and how to become a service leader in your industry.

Keywords: customer perspective, product marketing, service marketing, rigorous management

Procedia PDF Downloads 370

Authors: Erick C. Valverde, Wansu Lim

Abstract:

Facial emotion recognition (FER) system has been recently developed for more advanced computer vision applications. The ability to identify human emotions would enable smart healthcare facility to diagnose mental health illnesses (e.g., depression and stress) as well as better human social interactions with smart technologies. The FER system involves two steps: 1) face detection task and 2) facial emotion recognition task. It classifies the human expression in various categories such as angry, disgust, fear, happy, sad, surprise, and neutral. This system requires intensive research to address issues with human diversity, various unique human expressions, and variety of human facial features due to age differences. These issues generally affect the ability of the FER system to detect human emotions with high accuracy. Early stage of FER systems used simple supervised classification task algorithms like K-nearest neighbors (KNN) and artificial neural networks (ANN). These conventional FER systems have issues with low accuracy due to its inefficiency to extract significant features of several human emotions. To increase the accuracy of FER systems, deep learning (DL)-based methods, like convolutional neural networks (CNN), are proposed. These methods can find more complex features in the human face by means of the deeper connections within its architectures. However, the inference speed and computational costs of a DL-based FER system is often disregarded in exchange for higher accuracy results. To cope with this drawback, an optimized DL-based FER system is proposed in this study.An extreme version of Inception V3, known as Xception model, is leveraged by applying different network optimization methods. Specifically, network pruning and quantization are used to enable lower computational costs and reduce memory usage, respectively. To support low resource requirements, a 68-landmark face detector from Dlib is used in the early step of the FER system.Furthermore, a DL compiler is utilized to incorporate advanced optimization techniques to the Xception model to improve the inference speed of the FER system. In comparison to VGG-Net and ResNet50, the proposed optimized DL-based FER system experimentally demonstrates the objectives of the network optimization methods used. As a result, the proposed approach can be used to create an efficient and real-time FER system.

Keywords: deep learning, face detection, facial emotion recognition, network optimization methods

Procedia PDF Downloads 118