The current study had conducted the life cycle analysis (LCA) to assess the environmental impact of microalgal wastewater treatment via an integrated membrane bioreactor. The functional unit selected for this analysis was 1 kg of treated microalgal wastewater with contaminants eliminated by ultrafiltration membrane fabricated from recycled polyethylene terephthalate waste. Meanwhile, the applied system boundary in this study was distinguished based on two scenarios, namely, cradle-to-gate encompassed wastewater treatment only and cradle-to-cradle which included the reutilization of treated wastewater to cultivate microalgae again. The environmental impacts and hotspots associated with the different stages of the wastewater treatment process had clearly elucidated that membrane treatment had ensued the highest impact, followed by microalgal harvesting, and finally cultivation. Among the environmental impact categories, water-related impact was found to be prominent in the following series: freshwater ecotoxicity, freshwater eutrophication and marine ecotoxicity. Notably, the key performance indicator of all environmental impact, i.e., the global warming potential was found to be very much lower at 2.94 × 10-4 kg CO2 eq as opposed to other literatures reported on the LCA of wastewater treatments using membranes. Overall, this study had proffered insights into the environmental impact of microalgal wastewater treatment and its stimulus for sustainable wastewater management. The findings of this study can be instrumental in making informed decision for optimizing microalgal wastewater treatment and reutilization assisted by membrane technology with an ultimate goal of enhancing sustainability.
With the advancement of technologies and growth of the economy, it is inevitable that more complex processes are deployed, producing more heterogeneous wastewater that comes from biomedical, biochemical and various biotechnological industries. While the conventional way of wastewater treatment could effectively reduce the chemical oxygen demand, pH and turbidity of wastewater, trace pollutants, specifically the endocrine disruptor compounds (EDCs) that exist in µg L-1 or ng L-1 have further hardened the detection and removal of these biochemical pollutants. Even in small amounts, EDC could interfere human's hormone, causing severe implications on human body. Hence, this review elucidates the recent insights regarding the effectiveness of an advanced 2D material based on titanium carbide (Ti3C2Tx), also known as MXene, in detecting and removing EDCs. MXene's highly tunable feature also allows its surface chemistry to be adjusted by adding chemicals with different functional groups to adsorb different kinds of EDCs for biochemical pollution mitigation. At the same time, the incorporation of MXene into sample matrices also further eases the analysis of trace pollutants down to ng L-1 levels, thereby making way for a more cleaner and comprehensive wastewater treatment. In that sense, this review also highlights the progress in synthesizing MXene from the conventional method to the more modern approaches, together with their respective key parameters. To further understand and attest to the efficacy of MXene, the limitations and current gaps of this potential agent are also accentuated, targeting to seek resolutions for a more sustainable application.
The quest for energy efficiency (EE) in multi-tier Heterogeneous Networks (HetNets) is observed within the context of surging high-speed data demands and the rapid proliferation of wireless devices. The analysis of existing literature underscores the need for more comprehensive strategies to realize genuinely energy-efficient HetNets. This research work contributes significantly by employing a systematic methodology, utilizing This model facilitates the assessment of network performance by considering the spatial distribution of network elements. The stochastic nature of the PPP allows for a realistic representation of the random spatial deployment of base stations and users in multi-tier HetNets. Additionally, an analytical framework for Quality of Service (QoS) provision based on D-DOSS simplifies the understanding of user-base station relationships and offers essential performance metrics. Moreover, an optimization problem formulation, considering coverage, energy maximization, and delay minimization constraints, aims to strike a balance between key network attributes. This research not only addresses crucial challenges in creating EE HetNets but also lays a foundation for future advancements in wireless network design, operation, and management, ultimately benefiting network operators and end-users alike amidst the growing demand for high-speed data and the increasing prevalence of wireless devices. The proposed D-DOSS approach not only offers insights for the systematic design and analysis of EE HetNets but also systematically outperforms other state-of-the-art techniques presented. The improvement in energy efficiency systematically ranges from 67% (min side) to 98% (max side), systematically demonstrating the effectiveness of the proposed strategy in achieving higher energy efficiency compared to existing strategies. This systematic research work establishes a strong foundation for the systematic evolution of energy-efficient HetNets. The systematic methodology employed ensures a comprehensive understanding of the complex interplay of network dynamics and user requirements in a multi-tiered environment.
Micro, Small, and Medium-sized Enterprises (MSMEs) in Thailand were assessed in this study to determine the short-term and long-term economic effects of post-COVID- 19 -, with the goal of developing policy guidelines that focus on the methods and strategies that will further develop and help recover these sectors. MSMEs are the most vulnerable and require assistants to combat the pandemic. This study assesses the perspectives of stakeholders on the development of mechanisms and the strategies applied to support vulnerable groups in Thailand, which mostly consist of women and children. The main data collection was gathered through online questionnaires that were distributed to various stakeholder groups. The tools used for analysis were advanced quantitative analysis tools that aid in achieving this research study's objectives, and data was examined primarily through the usage of path modeling, structural equation modeling (SEM), and descriptive analysis was among the methods used. The findings reveal that in the short term, MSMEs' ability to respond to COVID-19 implications has a significant impact on both financial and non-financial performance. Non-financial performance, on the other hand, is more affected by adaptability than financial performance. Demand shock from lockdowns and other COVID-19 cautionary interventions has a negative and significant impact on MSMEs' adaptability, financial performance, and non-financial performance. The demand shocks increased the vulnerability of MSMEs significantly but it was found that proper management of demand shock has helped stabilized and improve MSMEs' financial and non-financial performances, as well as helped decrease their vulnerability. When it comes to government policy, the focus is usually on enhancing the flexibility and financial performance of MSMEs. The government's legislative actions have little impact on MSMEs' non-financial performance and vulnerability. This could be because the majority of the programs are more focused on providing financial assistance to businesses or their consumers. COVID-19's supply and demand shock only hindered MSMEs' ability to respond to the changes and challenges caused by the pandemic, according to vendors. The vulnerability of MSMEs caused by COVID-19 creates grave effects on their financial performance. The findings of this research paper will assist policymakers in identifying the most vulnerable aspects of MSMEs, as well as their expectations- and determine the forms of support that will be required to combat the current and future pandemic situations that may occur in Thailand. In addition, it will aid policymakers in the establishment of procedures and supporting strategies for MSMEs to reduce the unemployment rate and stimulate the Thai economy, among other factors of improvement.
Sustainable energy transition has brought the attention towards microalgae utilization as potential feedstock due to its tremendous capabilities over its predecessors for generating more energy with reduced carbon footprint. However, the commercialization of microalgae feedstock remains debatable due to the various factors and considerations taken into scaling-up the conventional microalgal upstream processes. This review provides a state-of-the-art assessment over the recent developments of available and existing microalgal upstream cultivation systems catered for maximum biomass production. The key growth parameters and main cultivation modes necessary for optimized microalgal growth conditions along with the fundamental aspects were also reviewed and evaluated comprehensively. In addition, the advancements and strategies towards potential scale-up of the microalgal cultivation technologies were highlighted to provide insights for further development into the upstream processes aimed at sustainable circular bioeconomy.
Perfluoroalkyl carboxylic acids (PFCAs) are sub-class of perfluoroalkyl substances commonly detected in water matrices. They are persistent in the environment, hence highly toxic to living organisms. Their occurrence at trace amount, complex nature and prone to matrix interference make their extraction and detection a challenge. This study consolidates current advancements in solid-phase extraction (SPE) techniques for the trace-level analysis of PFCAs from water matrices. The advantages of the methods in terms of ease of applications, low-cost, robustness, low solvents consumption, high pre-concentration factors, better extraction efficiency, good selectivity and recovery of the analytes have been emphasized. The article also demonstrated effectiveness of some porous materials for the adsorptive removal of the PFCAs from the water matrices. Mechanisms of the SPE/adsorption techniques have been discussed. The success and limitations of the processes have been elucidated.
We investigated the macroeconomic determinants of neonatal, infant, and under-five mortalities in Bangladesh for the period 1991-2018 and discuss implications of the United Nations' Sustainable Development Goal 3 (SDG 3) and Millennium Development Goal 4 (MDG 4) for developing countries. We used annual time series data and the econometric techniques of Fully Modified Ordinary Least Squares (FMOLS) and Dynamic Ordinary Least Squares (DOLS) regressions for analysis. Determinants most effective in combating neonatal, infant, and under-five mortalities include variables such as 'protecting newborns against tetanus', 'increasing healthcare expenditure', and 'making sure births are attended by skilled healthcare staff'. Employing more healthcare workers and assuring more and improved healthcare provisions can further reduce the neonatal, infant, and under-five mortalities. Developing countries with similar macroeconomic profiles can achieve similar SDG 3 and MDG 4 outcomes by emulating the policies and strategies Bangladesh applied to reducing child mortalities over the last three decades.
Cloud computing (CC) benefits and opportunities are among the fastest growing technologies in the computer industry. Cloud computing's challenges include resource allocation, security, quality of service, availability, privacy, data management, performance compatibility, and fault tolerance. Fault tolerance (FT) refers to a system's ability to continue performing its intended task in the presence of defects. Fault-tolerance challenges include heterogeneity and a lack of standards, the need for automation, cloud downtime reliability, consideration for recovery point objects, recovery time objects, and cloud workload. The proposed research includes machine learning (ML) algorithms such as naïve Bayes (NB), library support vector machine (LibSVM), multinomial logistic regression (MLR), sequential minimal optimization (SMO), K-nearest neighbor (KNN), and random forest (RF) as well as a fault-tolerance method known as delta-checkpointing to achieve higher accuracy, lesser fault prediction error, and reliability. Furthermore, the secondary data were collected from the homonymous, experimental high-performance computing (HPC) system at the Swiss Federal Institute of Technology (ETH), Zurich, and the primary data were generated using virtual machines (VMs) to select the best machine learning classifier. In this article, the secondary and primary data were divided into two split ratios of 80/20 and 70/30, respectively, and cross-validation (5-fold) was used to identify more accuracy and less prediction of faults in terms of true, false, repair, and failure of virtual machines. Secondary data results show that naïve Bayes performed exceptionally well on CPU-Mem mono and multi blocks, and sequential minimal optimization performed very well on HDD mono and multi blocks in terms of accuracy and fault prediction. In the case of greater accuracy and less fault prediction, primary data results revealed that random forest performed very well in terms of accuracy and fault prediction but not with good time complexity. Sequential minimal optimization has good time complexity with minor differences in random forest accuracy and fault prediction. We decided to modify sequential minimal optimization. Finally, the modified sequential minimal optimization (MSMO) algorithm with the fault-tolerance delta-checkpointing (D-CP) method is proposed to improve accuracy, fault prediction error, and reliability in cloud computing.
LoRa is an ISM-band based LPWAN communication protocol. Despite their wide network penetration of approximately 20 kilometers or higher using lower than 14 decibels transmitting power, it has been extensively documented and used in academia and industry. Although LoRa connectivity defines a public platform and enables users to create independent low-power wireless connections while relying on external architecture, it has gained considerable interest from scholars and the market. The two fundamental components of this platform are LoRaWAN and LoRa PHY. The consumer LoRaWAN component of the technology describes the network model, connectivity procedures, ability to operate the frequency range, and the types of interlinked gadgets. In contrast, the LoRa PHY component is patentable and provides information on the modulation strategy which is being utilized and its attributes. There are now several LoRa platforms available. To create usable LoRa systems, there are presently several technical difficulties to be overcome, such as connection management, allocation of resources, consistent communications, and security. This study presents a thorough overview of LoRa networking, covering the technological difficulties in setting up LoRa infrastructures and current solutions. Several outstanding challenges of LoRa communication are presented depending on our thorough research of the available solutions. The research report aims to stimulate additional research toward enhancing the LoRa Network capacity and allowing more realistic installations.
The benefits and opportunities offered by cloud computing are among the fastest-growing technologies in the computer industry. Additionally, it addresses the difficulties and issues that make more users more likely to accept and use the technology. The proposed research comprised of machine learning (ML) algorithms is Naïve Bayes (NB), Library Support Vector Machine (LibSVM), Multinomial Logistic Regression (MLR), Sequential Minimal Optimization (SMO), K Nearest Neighbor (KNN), and Random Forest (RF) to compare the classifier gives better results in accuracy and less fault prediction. In this research, the secondary data results (CPU-Mem Mono) give the highest percentage of accuracy and less fault prediction on the NB classifier in terms of 80/20 (77.01%), 70/30 (76.05%), and 5 folds cross-validation (74.88%), and (CPU-Mem Multi) in terms of 80/20 (89.72%), 70/30 (90.28%), and 5 folds cross-validation (92.83%). Furthermore, on (HDD Mono) the SMO classifier gives the highest percentage of accuracy and less fault prediction fault in terms of 80/20 (87.72%), 70/30 (89.41%), and 5 folds cross-validation (88.38%), and (HDD-Multi) in terms of 80/20 (93.64%), 70/30 (90.91%), and 5 folds cross-validation (88.20%). Whereas, primary data results found RF classifier gives the highest percentage of accuracy and less fault prediction in terms of 80/20 (97.14%), 70/30 (96.19%), and 5 folds cross-validation (95.85%) in the primary data results, but the algorithm complexity (0.17 seconds) is not good. In terms of 80/20 (95.71%), 70/30 (95.71%), and 5 folds cross-validation (95.71%), SMO has the second highest accuracy and less fault prediction, but the algorithm complexity is good (0.3 seconds). The difference in accuracy and less fault prediction between RF and SMO is only (.13%), and the difference in time complexity is (14 seconds). We have decided that we will modify SMO. Finally, the Modified Sequential Minimal Optimization (MSMO) Algorithm method has been proposed to get the highest accuracy & less fault prediction errors in terms of 80/20 (96.42%), 70/30 (96.42%), & 5 fold cross validation (96.50%).
This study examines the impact of artificial intelligence (AI) on loss in decision-making, laziness, and privacy concerns among university students in Pakistan and China. Like other sectors, education also adopts AI technologies to address modern-day challenges. AI investment will grow to USD 253.82 million from 2021 to 2025. However, worryingly, researchers and institutions across the globe are praising the positive role of AI but ignoring its concerns. This study is based on qualitative methodology using PLS-Smart for the data analysis. Primary data was collected from 285 students from different universities in Pakistan and China. The purposive Sampling technique was used to draw the sample from the population. The data analysis findings show that AI significantly impacts the loss of human decision-making and makes humans lazy. It also impacts security and privacy. The findings show that 68.9% of laziness in humans, 68.6% in personal privacy and security issues, and 27.7% in the loss of decision-making are due to the impact of artificial intelligence in Pakistani and Chinese society. From this, it was observed that human laziness is the most affected area due to AI. However, this study argues that significant preventive measures are necessary before implementing AI technology in education. Accepting AI without addressing the major human concerns would be like summoning the devils. Concentrating on justified designing and deploying and using AI for education is recommended to address the issue.
The findings of an extensive experimental research study on the usage of nano-sized cement powder and other additives combined to form cement-fine-aggregate matrices are discussed in this work. In the laboratory, dry and wet methods were used to create nano-sized cements. The influence of these nano-sized cements, nano-silica fumes, and nano-fly ash in different proportions was studied to the evaluate the engineering properties of the cement-fine-aggregate matrices concerning normal-sized, commercially available cement. The composites produced with modified cement-fine-aggregate matrices were subjected to microscopic-scale analyses using a petrographic microscope, a Scanning Electron Microscope (SEM), and a Transmission Electron Microscope (TEM). These studies unravelled the placement and behaviour of additives in controlling the engineering properties of the mix. The test results indicated that nano-cement and nano-sized particles improved the engineering properties of the hardened cement matrix. The wet-ground nano-cement showed the best result, 40 MPa 28th-day compressive strength, without mixing any additive compared with ordinary and dry-ground cements. The mix containing 50:50 normal and wet-ground cement exhibited 37.20 MPa 28th-day compressive strength. All other mixes with nano-sized dry cement, silica fume, and fly ash with different permutations and combinations gave better results than the normal-cement-fine-aggregate mix. The petrographic studies and the Scanning Electron Microscope (SEM) and Transmission Electron Microscope (TEM) analyses further validated the above findings. Statistical analyses and techniques such as correlation and stepwise multiple regression analysis were conducted to compose a predictive equation to calculate the 28th-day compressive strength. In addition to these methods, a repeated measures Analysis of Variance (ANOVA) was also implemented to analyse the statistically significant differences among three differently timed strength readings.
Rice is one of the vital foods consumed in most countries throughout the world. To estimate the yield, crop counting is used to indicate improper growth, identification of loam land, and control of weeds. It is becoming necessary to grow crops healthy, precisely, and proficiently as the demand increases for food supplies. Traditional counting methods have numerous disadvantages, such as long delay times and high sensitivity, and they are easily disturbed by noise. In this research, the detection and counting of rice plants using an unmanned aerial vehicle (UAV) and aerial images with a geographic information system (GIS) are used. The technique is implemented in the area of forty acres of rice crop in Tando Adam, Sindh, Pakistan. To validate the performance of the proposed system, the obtained results are compared with the standard plant count techniques as well as approved by the agronomist after testing soil and monitoring the rice crop count in each acre of land of rice crops. From the results, it is found that the proposed system is precise and detects rice crops accurately, differentiates from other objects, and estimates the soil health based on plant counting data; however, in the case of clusters, the counting is performed in semi-automated mode.
Cardiovascular disease includes coronary artery diseases (CAD), which include angina and myocardial infarction (commonly known as a heart attack), and coronary heart diseases (CHD), which are marked by the buildup of a waxy material called plaque inside the coronary arteries. Heart attacks are still the main cause of death worldwide, and if not treated right they have the potential to cause major health problems, such as diabetes. If ignored, diabetes can result in a variety of health problems, including heart disease, stroke, blindness, and kidney failure. Machine learning methods can be used to identify and diagnose diabetes and other illnesses. Diabetes and cardiovascular disease both can be diagnosed using several classifier types. Naive Bayes, K-Nearest neighbor (KNN), linear regression, decision trees (DT), and support vector machines (SVM) were among the classifiers employed, although all of these models had poor accuracy. Therefore, due to a lack of significant effort and poor accuracy, new research is required to diagnose diabetes and cardiovascular disease. This study developed an ensemble approach called "Stacking Classifier" in order to improve the performance of integrated flexible individual classifiers and decrease the likelihood of misclassifying a single instance. Naive Bayes, KNN, Linear Discriminant Analysis (LDA), and Decision Tree (DT) are just a few of the classifiers used in this study. As a meta-classifier, Random Forest and SVM are used. The suggested stacking classifier obtains a superior accuracy of 0.9735 percent when compared to current models for diagnosing diabetes, such as Naive Bayes, KNN, DT, and LDA, which are 0.7646 percent, 0.7460 percent, 0.7857 percent, and 0.7735 percent, respectively. Furthermore, for cardiovascular disease, when compared to current models such as KNN, NB, DT, LDA, and SVM, which are 0.8377 percent, 0.8256 percent, 0.8426 percent, 0.8523 percent, and 0.8472 percent, respectively, the suggested stacking classifier performed better and obtained a higher accuracy of 0.8871 percent.
We isolated a novel coronavirus from a medical team member presenting with fever and malaise after travel to Haiti. The virus showed 99.4% similarity with a recombinant canine coronavirus recently identified in a pneumonia patient in Malaysia, suggesting that infection with this virus and/or recombinant variants occurs in multiple locations.
Pandemics leave their mark quickly. This is true for all pandemics, including COVID-19. Its multifarious presence has wreaked havoc on people's physical, economic, and social life since late 2019. Despite the need for social science to save lives, it is also critical to ensure future generations are protected. COVID-19 appeared as the world grappled with the epidemic of climate change. This study suggests policymakers and practitioners address climate change and COVID-19 together. This article offers a narrative review of both pandemics' impacts. Scopus and Web of Science were sought databases. The findings are reported analytically using important works of contemporary social theorists. The analysis focuses on three interconnected themes: technology advancements have harmed vulnerable people; pandemics have macro- and micro-dimensions; and structural disparities. To conclude, we believe that collaborative effort is the key to combating COVID-19 and climate change, while understanding the lessons learnt from the industrialised world. Finally, policymakers can decrease the impact of global catastrophes by addressing many socioeconomic concerns concurrently.
Boron nitride (BN) nanomaterials are rapidly being investigated for potential applications in biomedical sciences due to their exceptional physico-chemical characteristics. However, their safe use demands a thorough understanding of their possible environmental and toxicological effects. The cytotoxicity of boron nitride nanotubes (BNNTs) was explored to see if they could be used in living cell imaging. It was observed that the cytotoxicity of BNNTs is higher in cancer cells (65 and 80%) than in normal cell lines (40 and 60%) for 24 h and 48 h respectively. The influence of multiple experimental parameters such as pH, time, amount of catalyst, and initial dye concentration on percentage degradation efficiency was also examined for both catalyst and dye. The degradation effectiveness decreases (92 to 25%) as the original concentration of dye increases (5-50 ppm) due to a decrease in the availability of adsorption sites. Similarly, the degradation efficiency improves up to 90% as the concentration of catalyst increases (0.01-0.05 g) due to an increase in the adsorption sites. The influence of pH was also investigated, the highest degradation efficiency for MO dye was observed at pH 4. Our results show that lower concentrations of BNNTs can be employed in biomedical applications. Dye degradation properties of BNNTs suggest that it can be a potential candidate as a wastewater and air treatment material.
Reduced electricity demand through the implementation of an energy efficiency policy is a central pillar of the Malaysian government's energy strategy. Energy efficiency first emerged as part of Malaysia's energy policy agenda in 1979 but only came into force during the 2000s. Initially, it was seen from global fears about the shortage of fossil fuels, then as a way of combating climate change. This paper offers a comprehensive review of Malaysia's energy policies with a focus on adopting policies to improve energy efficiency. Starting with Malaysia's preliminary policy in response to the OPEC-driven global oil crisis in 1973, the paper discusses how policymakers are considering energy efficiency from Malaysia's sustainable development perspective and what relevant government efforts have been made to improve it. The review evaluates the progress that has been made over the past 25 years to address energy efficiency in the economy and highlights the achievements and remaining difficulties. Findings show that the level of energy efficiency while having shown improvement during 1990-2015 was lower than expected. In terms of electricity intensity of GDP, Malaysia has a relatively large position among the ASEAN countries and the world's largest electricity consumers. Researchers, scientists, and practitioners will benefit from the extensive review material of this study, which will help them better understand energy efficiency and the sustainability strategy implemented in Malaysia to date.
Cancer is one of the most prevalent diseases worldwide. The most prevalent condition in women when aberrant cells develop out of control is breast cancer. Breast cancer detection and classification are exceedingly difficult tasks. As a result, several computational techniques, including k-nearest neighbor (KNN), support vector machine (SVM), multilayer perceptron (MLP), decision tree (DT), and genetic algorithms, have been applied in the current computing world for the diagnosis and classification of breast cancer. However, each method has its own limitations to how accurately it can be utilized. A novel convolutional neural network (CNN) model based on the Visual Geometry Group network (VGGNet) was also suggested in this study. The 16 layers in the current VGGNet-16 model lead to overfitting on the training and test data. We, thus, propose the VGGNet-12 model for breast cancer classification. The VGGNet-16 model has the problem of overfitting the breast cancer classification dataset. Based on the overfitting issues in the existing model, this research reduced the number of different layers in the VGGNet-16 model to solve the overfitting problem in this model. Because various models of the VGGNet, such as VGGNet-13 and VGGNet-19, were developed, this study proposed a new version of the VGGNet model, that is, the VGGNet-12 model. The performance of this model is checked using the breast cancer dataset, as compared to the CNN and LeNet models. From the simulation result, it can be seen that the proposed VGGNet-12 model enhances the simulation result as compared to the model used in this study. Overall, the experimental findings indicate that the suggested VGGNet-12 model did well in classifying breast cancer in terms of several characteristics.
Deep neural networks have made tremendous strides in the categorization of facial photos in the last several years. Due to the complexity of features, the enormous size of the picture/frame, and the severe inhomogeneity of image data, efficient face image classification using deep convolutional neural networks remains a challenge. Therefore, as data volumes continue to grow, the effective categorization of face photos in a mobile context utilizing advanced deep learning techniques is becoming increasingly important. In the recent past, some Deep Learning (DL) approaches for learning to identify face images have been designed; many of them use convolutional neural networks (CNNs). To address the problem of face mask recognition in facial images, we propose to use a Depthwise Separable Convolution Neural Network based on MobileNet (DWS-based MobileNet). The proposed network utilizes depth-wise separable convolution layers instead of 2D convolution layers. With limited datasets, the DWS-based MobileNet performs exceptionally well. DWS-based MobileNet decreases the number of trainable parameters while enhancing learning performance by adopting a lightweight network. Our technique outperformed the existing state of the art when tested on benchmark datasets. When compared to Full Convolution MobileNet and baseline methods, the results of this study reveal that adopting Depthwise Separable Convolution-based MobileNet significantly improves performance (Acc. = 93.14, Pre. = 92, recall = 92, F-score = 92).