Pub Date : 2023-07-01DOI: 10.58245/ipsi.tir.2302.09
G. Savić, M. Segedinac, Milan Čeliković, I. Luković
We propose a software solution for representing diverse sets of key performance indicators in higher education. Our solution addresses both the heterogeneity and the common structure of key performance indicators. To tackle the issue of heterogeneity, we employ metamodeling and propose a meta-model that is expressive and generic enough to represent any set of key performance indicators in higher education. The proposed meta-model is more abstract than any specific key performance indicators set, and the sets are considered as models, which are instances of the proposed metamodel. We address the heterogeneity in calculating the key performance indicators' values by representing them with mathematical formulas and utilizing an expression language that allows for their dynamic evaluation. We verified the solution by representing typical key performance indicator sets and developing a software application prototype that enables the creation, monitoring, and further development of key performance indicator sets. The verification confirms the wide applicability of our proposed solution.
{"title":"A Meta-model for Key Performance Indicators in Higher Education","authors":"G. Savić, M. Segedinac, Milan Čeliković, I. Luković","doi":"10.58245/ipsi.tir.2302.09","DOIUrl":"https://doi.org/10.58245/ipsi.tir.2302.09","url":null,"abstract":"We propose a software solution for representing diverse sets of key performance indicators in higher education. Our solution addresses both the heterogeneity and the common structure of key performance indicators. To tackle the issue of heterogeneity, we employ metamodeling and propose a meta-model that is expressive and generic enough to represent any set of key performance indicators in higher education. The proposed meta-model is more abstract than any specific key performance indicators set, and the sets are considered as models, which are instances of the proposed metamodel. We address the heterogeneity in calculating the key performance indicators' values by representing them with mathematical formulas and utilizing an expression language that allows for their dynamic evaluation. We verified the solution by representing typical key performance indicator sets and developing a software application prototype that enables the creation, monitoring, and further development of key performance indicator sets. The verification confirms the wide applicability of our proposed solution.","PeriodicalId":41192,"journal":{"name":"IPSI BgD Transactions on Internet Research","volume":null,"pages":null},"PeriodicalIF":0.4,"publicationDate":"2023-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"88678498","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-07-01DOI: 10.58245/ipsi.tir.2302.10
Srđa Bjeladinović, Milica Škembarević, Olga Jejic, Marko Asanović,
JSON is a popular and proven standard for specifying self-describing text files with a flexible structure. To maintain its position in the market, Oracle introduced support for JSON data in the 12c R1 version of its DBMS. This version has introduced functions for storing and managing JSON data in native form but also showed some limitations. Each new version introduced new or updated JSON functions. The 21c can store JSON data in binary form, provides more straightforward syntax and even supports JSON as a predefined data type. The paper aims to compare the performance when the underlying storage of JSON is native or binary. A data model and seven use cases were designed to demonstrate earlier and new functionalities. Additionally, experiments showed the impact of JSON data stored in native (19c and 21c) and binary form (21c) on the average execution time and costs of SQL statements.
{"title":"An analysis of using binary JSON versus native JSON on the example of Oracle DBMS","authors":"Srđa Bjeladinović, Milica Škembarević, Olga Jejic, Marko Asanović,","doi":"10.58245/ipsi.tir.2302.10","DOIUrl":"https://doi.org/10.58245/ipsi.tir.2302.10","url":null,"abstract":"JSON is a popular and proven standard for specifying self-describing text files with a flexible structure. To maintain its position in the market, Oracle introduced support for JSON data in the 12c R1 version of its DBMS. This version has introduced functions for storing and managing JSON data in native form but also showed some limitations. Each new version introduced new or updated JSON functions. The 21c can store JSON data in binary form, provides more straightforward syntax and even supports JSON as a predefined data type. The paper aims to compare the performance when the underlying storage of JSON is native or binary. A data model and seven use cases were designed to demonstrate earlier and new functionalities. Additionally, experiments showed the impact of JSON data stored in native (19c and 21c) and binary form (21c) on the average execution time and costs of SQL statements.","PeriodicalId":41192,"journal":{"name":"IPSI BgD Transactions on Internet Research","volume":null,"pages":null},"PeriodicalIF":0.4,"publicationDate":"2023-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"88436995","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-07-01DOI: 10.58245/ipsi.tir.2302.02
Liangdong Deng, Arpan Mahara, N. Rishe, Malek Adjouadi
Geographic datasets are usually accompanied by spatial non-stationarity – a phenomenon that the relationship between features varies across space. Naturally, nonstationarity can be interpreted as the underlying rule that decides how data are generated and alters over space. Therefore, traditional machine learning algorithms are not suitable for handling non-stationary geographic datasets, as they only render a single global model. To solve this problem, researchers often adopt the multiple-local-model approach, which uses different models to account for different sub-regions of space. This approach has been proven efficient but not optimal, as it is inherently difficult to decide the size of subregions. Additionally, the fact that local models are only trained on a subset of data also limits their potential. This paper proposes an entirely different strategy that interprets nonstationarity as a lack of data and addresses it by introducing latent variables to the original dataset. Backpropagation is then used to find the best values for these latent variables. Experiments show that this method is at least as efficient as multiple-local-model-based approaches and has even greater potential.
{"title":"LVRF: A Latent Variable Based Approach for Exploring Geographic Datasets","authors":"Liangdong Deng, Arpan Mahara, N. Rishe, Malek Adjouadi","doi":"10.58245/ipsi.tir.2302.02","DOIUrl":"https://doi.org/10.58245/ipsi.tir.2302.02","url":null,"abstract":"Geographic datasets are usually accompanied by spatial non-stationarity – a phenomenon that the relationship between features varies across space. Naturally, nonstationarity can be interpreted as the underlying rule that decides how data are generated and alters over space. Therefore, traditional machine learning algorithms are not suitable for handling non-stationary geographic datasets, as they only render a single global model. To solve this problem, researchers often adopt the multiple-local-model approach, which uses different models to account for different sub-regions of space. This approach has been proven efficient but not optimal, as it is inherently difficult to decide the size of subregions. Additionally, the fact that local models are only trained on a subset of data also limits their potential. This paper proposes an entirely different strategy that interprets nonstationarity as a lack of data and addresses it by introducing latent variables to the original dataset. Backpropagation is then used to find the best values for these latent variables. Experiments show that this method is at least as efficient as multiple-local-model-based approaches and has even greater potential.","PeriodicalId":41192,"journal":{"name":"IPSI BgD Transactions on Internet Research","volume":null,"pages":null},"PeriodicalIF":0.4,"publicationDate":"2023-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"76594173","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-07-01DOI: 10.58245/ipsi.tir.2302.11
Jovan G. Radosavljević
Diameter-2-critical graphs (abbr. D2C) are diameter 2 graphs whose diameter increases by removing any edge. The procedure used to obtain the list of D2C graphs of the order at most 13 is described. This is achieved by incorporating the diameter 2 test and the criticality test into geng, the program from the package nauty that generates the list of all non-isomorphic connected graphs. Experiments with the two heuristics in diameter 2 test, which is intensively used during the search, show that it is slightly more efficient to start the test with the largest degree node using BFS algorithm. As an application of the obtained list, the three conjectures concerning the maximum number of edges in D2C graphs were checked for graphs of the order at most 13 and one counterexample was found. Index Terms: diameter-2-critical graphs, graph diameter, primitive graph.
{"title":"Diameter-2-critical graphs with at most 13 nodes","authors":"Jovan G. Radosavljević","doi":"10.58245/ipsi.tir.2302.11","DOIUrl":"https://doi.org/10.58245/ipsi.tir.2302.11","url":null,"abstract":"Diameter-2-critical graphs (abbr. D2C) are diameter 2 graphs whose diameter increases by removing any edge. The procedure used to obtain the list of D2C graphs of the order at most 13 is described. This is achieved by incorporating the diameter 2 test and the criticality test into geng, the program from the package nauty that generates the list of all non-isomorphic connected graphs. Experiments with the two heuristics in diameter 2 test, which is intensively used during the search, show that it is slightly more efficient to start the test with the largest degree node using BFS algorithm. As an application of the obtained list, the three conjectures concerning the maximum number of edges in D2C graphs were checked for graphs of the order at most 13 and one counterexample was found. Index Terms: diameter-2-critical graphs, graph diameter, primitive graph.","PeriodicalId":41192,"journal":{"name":"IPSI BgD Transactions on Internet Research","volume":null,"pages":null},"PeriodicalIF":0.4,"publicationDate":"2023-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"90024705","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-07-01DOI: 10.58245/ipsi.tir.2302.07
Boris Delibasic, S. Radovanović, S. Vukanovic
This paper proposes a decision support system for internal migration policy in the Republic of Serbia, which uses machine learning and knowledge extraction methods to analyze data and identify key features for policy decision-making. Internal migration is an issue that creates uneven development and sustainability challenges in countries. More specifically, internal migrations are putting a big pressure on cities and urban areas, while leaving vast less-urbanized areas depopulated and unsustainable to future generations. This paper includes two machine learning models with an accuracy of 70% for predicting internal migration intensity in local selfgovernments (LSGs), as well as the proposed decision-support tool that achieves an accuracy of 66%. The proposed system maintains desirable properties of decision support systems such as correctness, completeness, consistency, comprehensibility, and convenience and allows the what-if analysis to evaluate appropriate policies for each LSG. The identified key features can be used to influence migration levels in LSGs and promote balanced development in Serbia.
{"title":"A Decision Support System for Internal Migration Policy-Making","authors":"Boris Delibasic, S. Radovanović, S. Vukanovic","doi":"10.58245/ipsi.tir.2302.07","DOIUrl":"https://doi.org/10.58245/ipsi.tir.2302.07","url":null,"abstract":"This paper proposes a decision support system for internal migration policy in the Republic of Serbia, which uses machine learning and knowledge extraction methods to analyze data and identify key features for policy decision-making. Internal migration is an issue that creates uneven development and sustainability challenges in countries. More specifically, internal migrations are putting a big pressure on cities and urban areas, while leaving vast less-urbanized areas depopulated and unsustainable to future generations. This paper includes two machine learning models with an accuracy of 70% for predicting internal migration intensity in local selfgovernments (LSGs), as well as the proposed decision-support tool that achieves an accuracy of 66%. The proposed system maintains desirable properties of decision support systems such as correctness, completeness, consistency, comprehensibility, and convenience and allows the what-if analysis to evaluate appropriate policies for each LSG. The identified key features can be used to influence migration levels in LSGs and promote balanced development in Serbia.","PeriodicalId":41192,"journal":{"name":"IPSI BgD Transactions on Internet Research","volume":null,"pages":null},"PeriodicalIF":0.4,"publicationDate":"2023-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"84766618","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-07-01DOI: 10.58245/ipsi.tir.2302.05
N. Rishe, Dan Tamir, Malek Adjouadi
resented here is a model objectivizing real estate prices so that prices across time could be compared to understand historical price trends and also to assist in a property evaluation or appraisal, as well as for the analysis of comparables in estimating a reasonable offer for a property on the market. Given a timespan of interest, a locale (e.g., a particular zipcode, a city, a county, a state), a category of properties of interest (e.g., condos), an objective historical trend in values can be computed by first evaluating the ratios between the transactions’ realized prices and objective governmental assessment of the properties at some fixed point of time; then, for each period (a month) averaging the ratios of all transaction in that period; then, comparing said averages (or medians) between different periods.
{"title":"Spatiotemporal Model of Real Estate Valuation Trend","authors":"N. Rishe, Dan Tamir, Malek Adjouadi","doi":"10.58245/ipsi.tir.2302.05","DOIUrl":"https://doi.org/10.58245/ipsi.tir.2302.05","url":null,"abstract":"resented here is a model objectivizing real estate prices so that prices across time could be compared to understand historical price trends and also to assist in a property evaluation or appraisal, as well as for the analysis of comparables in estimating a reasonable offer for a property on the market. Given a timespan of interest, a locale (e.g., a particular zipcode, a city, a county, a state), a category of properties of interest (e.g., condos), an objective historical trend in values can be computed by first evaluating the ratios between the transactions’ realized prices and objective governmental assessment of the properties at some fixed point of time; then, for each period (a month) averaging the ratios of all transaction in that period; then, comparing said averages (or medians) between different periods.","PeriodicalId":41192,"journal":{"name":"IPSI BgD Transactions on Internet Research","volume":null,"pages":null},"PeriodicalIF":0.4,"publicationDate":"2023-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"89593595","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-07-01DOI: 10.58245/ipsi.tir.2302.08
Dajana Antanasijević, Marko Vještica, Vladimir Dimitrieski, L. Grubić-Nešić, S. Ristić, M. Pisarić
Although Industry 4.0 improved humanmachine relationship in technical aspects, it failed to put human needs at the focus of the production process. Industry 5.0 is complementing the Industry 4.0 focusing on the workers’ skills, knowledge, and abilities to cooperate with machines and robots. In our previous research, we proposed a framework for the formal description and automatic execution of production processes within Industry 4.0. As a result, a Domain-Specific Modeling Language (DSML) named Multi-Level Production Process Modeling Language (MultiProLan) was created aimed at modeling production processes at different levels of abstraction. The importance given to the workers within Industry 5.0 motivated us to investigate two different roles of a human worker: as an employee within an organization and as a human production resource. We propose a DSML named HResModLan aimed at human resource modeling from two different perspectives: organizational and production. The part of HResModLan language representing the organizational perspective is presented in this paper. The main goal of its creation is to enable the easier and more effective requiring, selection, hiring and development of employees within an organization. The paper presents an analysis of the human resource domain, abstract and concrete syntaxes of the HResModLan language, and a model of a furniture factory and its employees expressed using theconcepts of the HResModLan language.
{"title":"An Organizational Perspective of Human Resource Modeling","authors":"Dajana Antanasijević, Marko Vještica, Vladimir Dimitrieski, L. Grubić-Nešić, S. Ristić, M. Pisarić","doi":"10.58245/ipsi.tir.2302.08","DOIUrl":"https://doi.org/10.58245/ipsi.tir.2302.08","url":null,"abstract":"Although Industry 4.0 improved humanmachine relationship in technical aspects, it failed to put human needs at the focus of the production process. Industry 5.0 is complementing the Industry 4.0 focusing on the workers’ skills, knowledge, and abilities to cooperate with machines and robots. In our previous research, we proposed a framework for the formal description and automatic execution of production processes within Industry 4.0. As a result, a Domain-Specific Modeling Language (DSML) named Multi-Level Production Process Modeling Language (MultiProLan) was created aimed at modeling production processes at different levels of abstraction. The importance given to the workers within Industry 5.0 motivated us to investigate two different roles of a human worker: as an employee within an organization and as a human production resource. We propose a DSML named HResModLan aimed at human resource modeling from two different perspectives: organizational and production. The part of HResModLan language representing the organizational perspective is presented in this paper. The main goal of its creation is to enable the easier and more effective requiring, selection, hiring and development of employees within an organization. The paper presents an analysis of the human resource domain, abstract and concrete syntaxes of the HResModLan language, and a model of a furniture factory and its employees expressed using theconcepts of the HResModLan language.","PeriodicalId":41192,"journal":{"name":"IPSI BgD Transactions on Internet Research","volume":null,"pages":null},"PeriodicalIF":0.4,"publicationDate":"2023-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"82797619","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-07-01DOI: 10.58245/ipsi.tir.2302.03
Lawrence Egharevba, Sanjoy Kumar, N. Rishe, Hadi Amini, Malek Adjouadi
Deep Learning is becoming a very popular tool for generating and reconstructing images. Research has shown that deep learning algorithms can perform cutting-edge restoration tasks for various types of images. The performance of these algorithms can be achieved by training Deep Convolutional Neural Networks (DCNNs) with data from a large sample size. The processing of high-resolution satellite imagery becomes difficult when there are only a few images in a dataset. An approach based on the intrinsic properties of Deep Convolutional Neural Networks (DCNNs) is presented in this paper for the detection and removal of clouds from remote sensing images without any prior training. Our results demonstrated that the algorithm we used performed well when compared to trained algorithms.
{"title":"\"Detecting and Removing Clouds Affected Regions from Satellite Images Using Deep Learning\"","authors":"Lawrence Egharevba, Sanjoy Kumar, N. Rishe, Hadi Amini, Malek Adjouadi","doi":"10.58245/ipsi.tir.2302.03","DOIUrl":"https://doi.org/10.58245/ipsi.tir.2302.03","url":null,"abstract":"Deep Learning is becoming a very popular tool for generating and reconstructing images. Research has shown that deep learning algorithms can perform cutting-edge restoration tasks for various types of images. The performance of these algorithms can be achieved by training Deep Convolutional Neural Networks (DCNNs) with data from a large sample size. The processing of high-resolution satellite imagery becomes difficult when there are only a few images in a dataset. An approach based on the intrinsic properties of Deep Convolutional Neural Networks (DCNNs) is presented in this paper for the detection and removal of clouds from remote sensing images without any prior training. Our results demonstrated that the algorithm we used performed well when compared to trained algorithms.","PeriodicalId":41192,"journal":{"name":"IPSI BgD Transactions on Internet Research","volume":null,"pages":null},"PeriodicalIF":0.4,"publicationDate":"2023-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"88192527","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-07-01DOI: 10.58245/ipsi.tir.2302.06
Khandaker Mamun Ahmed, Farid Ghareh Mohammadi, M. Matus, Farzan Shenavarmasouleh, Luiz Manella Pereira, Zisis Ioannis, M. Amini
In the past few years, automatic building detection in aerial images has become an emerging field in computer vision. Detecting the specific types of houses will provide information in urbanization, change detection, and urban monitoring that play increasingly important roles in modern city planning and natural hazard preparedness. In this paper, we demonstrate the effectiveness of detecting various types of houses in aerial imagery using Faster Region-based Convolutional Neural Network (Faster-RCNN). After formulating the dataset and extracting bounding-box information, pre-trained ResNet50 is used to get the feature maps. The fully convolutional Region Proposal Network (RPN) first predicts the bounds and objectness score of objects (in this case house) from the feature maps. Then, the Region of Interest (RoI) pooling layer extracts interested regions to detect objects that are present in the images. To the best of our knowledge, this is the first attempt at detecting houses using Faster R-CNN that has achieved satisfactory results. This experiment opens a new path to conduct and extent the works not only for civil and environmental domain but also other applied science disciplines.
{"title":"\"Towards Real-time House Detection in Aerial Imagery Using Faster Region-based Convolutional Neural Network\"","authors":"Khandaker Mamun Ahmed, Farid Ghareh Mohammadi, M. Matus, Farzan Shenavarmasouleh, Luiz Manella Pereira, Zisis Ioannis, M. Amini","doi":"10.58245/ipsi.tir.2302.06","DOIUrl":"https://doi.org/10.58245/ipsi.tir.2302.06","url":null,"abstract":"In the past few years, automatic building detection in aerial images has become an emerging field in computer vision. Detecting the specific types of houses will provide information in urbanization, change detection, and urban monitoring that play increasingly important roles in modern city planning and natural hazard preparedness. In this paper, we demonstrate the effectiveness of detecting various types of houses in aerial imagery using Faster Region-based Convolutional Neural Network (Faster-RCNN). After formulating the dataset and extracting bounding-box information, pre-trained ResNet50 is used to get the feature maps. The fully convolutional Region Proposal Network (RPN) first predicts the bounds and objectness score of objects (in this case house) from the feature maps. Then, the Region of Interest (RoI) pooling layer extracts interested regions to detect objects that are present in the images. To the best of our knowledge, this is the first attempt at detecting houses using Faster R-CNN that has achieved satisfactory results. This experiment opens a new path to conduct and extent the works not only for civil and environmental domain but also other applied science disciplines.","PeriodicalId":41192,"journal":{"name":"IPSI BgD Transactions on Internet Research","volume":null,"pages":null},"PeriodicalIF":0.4,"publicationDate":"2023-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"86378569","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-07-01DOI: 10.58245/ipsi.tir.2302.04
Arpan Mahara, N. Rishe
In the past few years, there have been many research studies conducted in the field of Satellite Image Classification. The purposes of these studies included flood identification, forest fire monitoring, greenery land identification, and land-usage identification. In this field, finding suitable data is often considered problematic, and some research has also been done to identify and extract suitable datasets for classification. Although satellite data can be challenging to deal with, Convolutional Neural Networks (CNNs), which consist of multiple interconnected neurons, have shown promising results when applied to satellite imagery data. In the present work, first we have manually downloaded satellite images of four different classes in Florida locations using the TerraFly Mapping System, developed and managed by the High Performance Database Research Center at Florida International University. We then develop a CNN architecture suitable for extracting features and capable of multi-class classification in our dataset. We discuss the shortcomings in the classification due to the limited size of the dataset. To address this issue, we first employ data augmentation and then utilize transfer learning methodology for feature extraction with VGG16 and ResNet50 pretrained models. We use these features to classify satellite imagery of Florida. We analyze the misclassification in our model and, to address this issue, we introduce a location-based CNN model. We convert coordinates to geohash codes, use these codes as an additional feature vector and feed them into the CNN model. We believe that the new CNN model combined with geohash codes as location features provides a better accuracy for our dataset.
{"title":"\"Integrating Location Information as Geohash Codes in Convolutional Neural Network-Based Satellite Image Classification\"","authors":"Arpan Mahara, N. Rishe","doi":"10.58245/ipsi.tir.2302.04","DOIUrl":"https://doi.org/10.58245/ipsi.tir.2302.04","url":null,"abstract":"In the past few years, there have been many research studies conducted in the field of Satellite Image Classification. The purposes of these studies included flood identification, forest fire monitoring, greenery land identification, and land-usage identification. In this field, finding suitable data is often considered problematic, and some research has also been done to identify and extract suitable datasets for classification. Although satellite data can be challenging to deal with, Convolutional Neural Networks (CNNs), which consist of multiple interconnected neurons, have shown promising results when applied to satellite imagery data. In the present work, first we have manually downloaded satellite images of four different classes in Florida locations using the TerraFly Mapping System, developed and managed by the High Performance Database Research Center at Florida International University. We then develop a CNN architecture suitable for extracting features and capable of multi-class classification in our dataset. We discuss the shortcomings in the classification due to the limited size of the dataset. To address this issue, we first employ data augmentation and then utilize transfer learning methodology for feature extraction with VGG16 and ResNet50 pretrained models. We use these features to classify satellite imagery of Florida. We analyze the misclassification in our model and, to address this issue, we introduce a location-based CNN model. We convert coordinates to geohash codes, use these codes as an additional feature vector and feed them into the CNN model. We believe that the new CNN model combined with geohash codes as location features provides a better accuracy for our dataset.","PeriodicalId":41192,"journal":{"name":"IPSI BgD Transactions on Internet Research","volume":null,"pages":null},"PeriodicalIF":0.4,"publicationDate":"2023-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"91160878","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：