Pub Date : 2019-05-01DOI: 10.1109/ICCAIRO47923.2019.00029
D. Oreški, Goran Hajdin
This paper addresses the analysis of machine learning (ML) effectiveness in learning analytics context. Four different machine learning approaches are evaluated. The results offer information about the usefulness of these approaches and help to decide which of the approaches is the most promising one in learning analytics application. Results substantiate that the neural networks ML model trained on our learning management system (LMS) data exhibits the best performance for predicting the students' academic performance. In our future research, predictive model results will be explained within a pedagogical context in order to be used as part of student support mechanism.
{"title":"A Comparative Study of Machine Learning Approaches on Learning Management System Data","authors":"D. Oreški, Goran Hajdin","doi":"10.1109/ICCAIRO47923.2019.00029","DOIUrl":"https://doi.org/10.1109/ICCAIRO47923.2019.00029","url":null,"abstract":"This paper addresses the analysis of machine learning (ML) effectiveness in learning analytics context. Four different machine learning approaches are evaluated. The results offer information about the usefulness of these approaches and help to decide which of the approaches is the most promising one in learning analytics application. Results substantiate that the neural networks ML model trained on our learning management system (LMS) data exhibits the best performance for predicting the students' academic performance. In our future research, predictive model results will be explained within a pedagogical context in order to be used as part of student support mechanism.","PeriodicalId":297342,"journal":{"name":"2019 International Conference on Control, Artificial Intelligence, Robotics & Optimization (ICCAIRO)","volume":"187 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124746723","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 : 2019-05-01DOI: 10.1109/ICCAIRO47923.2019.00044
Orkény Zováthi, L. Kovács, Balázs Nagy, C. Benedek
In this paper we propose a novel approach for upgrading real time 3D dynamic object detection methods operating on rotating multi-beam (RMB) Lidar measurements using 3D background city maps stored in new generation geographic information systems (GIS) and previously detected dynamic objects propagated by tracking. First, we apply a state-of-the-art object detection method and distinguish the predicted dynamic object candidates and the remaining static regions of the current Lidar measurement. Next we find an optimal transformation between the static part of the RMB Lidar measurements and the background city map using a multimodal point cloud registration algorithm operating in the Hough space. After the accurate alignment, we filter false-positively detected object candidates in the RMB Lidar data based on the map. To find additional objects missed by the object detector on the current measurement, we apply a Kalman-filter based object tracking. Hereby we first predict the current state of the previously detected and tracked objects. Next, we apply a Hungarian matcher based assignment between the tracked and the current objects and update the object list according to the result. For better accuracy, we keep all predictions through a couple of frames. We evaluated our method qualitatively and quantitatively in crowded urban scenes of Budapest, Hungary, and the results showed that with background map based filtering we can achieve a 26,52% improvement detecting vehicles and 9,38% for pedestrians in precision, while via tracking, a 12,84% improvement for vehicles and 14,34% for pedestrians in recall against the state-of-the-art object detection method relying purely on a single Lidar time frame.
{"title":"Multi-Object Detection in Urban Scenes Utilizing 3D Background Maps and Tracking","authors":"Orkény Zováthi, L. Kovács, Balázs Nagy, C. Benedek","doi":"10.1109/ICCAIRO47923.2019.00044","DOIUrl":"https://doi.org/10.1109/ICCAIRO47923.2019.00044","url":null,"abstract":"In this paper we propose a novel approach for upgrading real time 3D dynamic object detection methods operating on rotating multi-beam (RMB) Lidar measurements using 3D background city maps stored in new generation geographic information systems (GIS) and previously detected dynamic objects propagated by tracking. First, we apply a state-of-the-art object detection method and distinguish the predicted dynamic object candidates and the remaining static regions of the current Lidar measurement. Next we find an optimal transformation between the static part of the RMB Lidar measurements and the background city map using a multimodal point cloud registration algorithm operating in the Hough space. After the accurate alignment, we filter false-positively detected object candidates in the RMB Lidar data based on the map. To find additional objects missed by the object detector on the current measurement, we apply a Kalman-filter based object tracking. Hereby we first predict the current state of the previously detected and tracked objects. Next, we apply a Hungarian matcher based assignment between the tracked and the current objects and update the object list according to the result. For better accuracy, we keep all predictions through a couple of frames. We evaluated our method qualitatively and quantitatively in crowded urban scenes of Budapest, Hungary, and the results showed that with background map based filtering we can achieve a 26,52% improvement detecting vehicles and 9,38% for pedestrians in precision, while via tracking, a 12,84% improvement for vehicles and 14,34% for pedestrians in recall against the state-of-the-art object detection method relying purely on a single Lidar time frame.","PeriodicalId":297342,"journal":{"name":"2019 International Conference on Control, Artificial Intelligence, Robotics & Optimization (ICCAIRO)","volume":"42 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126731914","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 : 2019-05-01DOI: 10.1109/ICCAIRO47923.2019.00028
Nebras Alattar, Jabar H. Yousif
The statistics of the World Health Organization (WHO) indicate that outdoor air pollution in 2016 is a significant cause of premature mortality, with an average of 4.2 million death cases. This mortality is due to exposure to PM2.5 particulate matter, which causes many diseases such as respiratory, cardiovascular, and cancers. The concentration of particulate matter (PM) is the most popular air pollutant that affects short term and long term health. The paper aims to study and investigate the concentration dispersion of particulates (PM 2.5 and PM10) and its impact on human health in Oman. The study suggested a hybrid neural and mathematical approaches for analyzing the effect rate of particulate matter (PM2.5 and PM10). The paper implements a comparative study to analyze the proposed neural and mathematical models, which predict the future levels of pollutants in a fast, cheap, and safe way. The Linear regression models achieve fewer results of R², MSE, RMSE (0.7604, 0.0673, 0.2595), respectively. However, the non-linear regression polynomial prediction model obtained excellent results based on the coefficient of determination (R²) value of 0.9394 and mean square error (MSE) rate of 0.0209, and root mean square error (RMSE) value of 0.1447. Moreover, the Neural SOM model obtained the highest results in predicting the experimental data that achieved an MSE value of 0.0064, correlation rate (R) value of 0.994, NMSE value of 0.01392, and MAE value of 0.0467. All the results were correctly verified based on suitable mathematical methods.
{"title":"Evaluating Particulate Matter (PM2.5 and PM10) Impact on Human Health in Oman Based on a Hybrid Artificial Neural Network and Mathematical Models","authors":"Nebras Alattar, Jabar H. Yousif","doi":"10.1109/ICCAIRO47923.2019.00028","DOIUrl":"https://doi.org/10.1109/ICCAIRO47923.2019.00028","url":null,"abstract":"The statistics of the World Health Organization (WHO) indicate that outdoor air pollution in 2016 is a significant cause of premature mortality, with an average of 4.2 million death cases. This mortality is due to exposure to PM2.5 particulate matter, which causes many diseases such as respiratory, cardiovascular, and cancers. The concentration of particulate matter (PM) is the most popular air pollutant that affects short term and long term health. The paper aims to study and investigate the concentration dispersion of particulates (PM 2.5 and PM10) and its impact on human health in Oman. The study suggested a hybrid neural and mathematical approaches for analyzing the effect rate of particulate matter (PM2.5 and PM10). The paper implements a comparative study to analyze the proposed neural and mathematical models, which predict the future levels of pollutants in a fast, cheap, and safe way. The Linear regression models achieve fewer results of R², MSE, RMSE (0.7604, 0.0673, 0.2595), respectively. However, the non-linear regression polynomial prediction model obtained excellent results based on the coefficient of determination (R²) value of 0.9394 and mean square error (MSE) rate of 0.0209, and root mean square error (RMSE) value of 0.1447. Moreover, the Neural SOM model obtained the highest results in predicting the experimental data that achieved an MSE value of 0.0064, correlation rate (R) value of 0.994, NMSE value of 0.01392, and MAE value of 0.0467. All the results were correctly verified based on suitable mathematical methods.","PeriodicalId":297342,"journal":{"name":"2019 International Conference on Control, Artificial Intelligence, Robotics & Optimization (ICCAIRO)","volume":"45 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133430207","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 : 2019-05-01DOI: 10.1109/ICCAIRO47923.2019.00040
D. Tarasov, O. Milder
In printing art, color management is largely based on the adjustment and analysis of the behavior of gradation curves calculated for the initial color channels. However, this approach does not take into account the mutual influence of colors and the change in hue, for example, when inks overlap. The approach proposed by the authors earlier offers to replace the two-dimensional tone reproduction curves with three-dimensional gradation trajectories in the CIE Lab metric space. In this paper, we develop this approach. It is shown that calculations using the mathematical apparatus of the differential geometry of spatial curves describing gradation trajectories might be simplified using the discrete approach. Discretization is associated with the peculiarities of color formation in modern digital printing systems. These features are used in the approximation of gradation trajectories using polynomials. In this case, color coordinates are considered as continuous functions of filling a discrete raster cell with dye. The proposed method allows one to calculate trajectories faster and without the use of cumbersome computations. An experimental verification of this approach was carried out using the example of a digital electrophotographic printing system.
{"title":"Discrete Gradation Trajectories Computation in Electrophotography","authors":"D. Tarasov, O. Milder","doi":"10.1109/ICCAIRO47923.2019.00040","DOIUrl":"https://doi.org/10.1109/ICCAIRO47923.2019.00040","url":null,"abstract":"In printing art, color management is largely based on the adjustment and analysis of the behavior of gradation curves calculated for the initial color channels. However, this approach does not take into account the mutual influence of colors and the change in hue, for example, when inks overlap. The approach proposed by the authors earlier offers to replace the two-dimensional tone reproduction curves with three-dimensional gradation trajectories in the CIE Lab metric space. In this paper, we develop this approach. It is shown that calculations using the mathematical apparatus of the differential geometry of spatial curves describing gradation trajectories might be simplified using the discrete approach. Discretization is associated with the peculiarities of color formation in modern digital printing systems. These features are used in the approximation of gradation trajectories using polynomials. In this case, color coordinates are considered as continuous functions of filling a discrete raster cell with dye. The proposed method allows one to calculate trajectories faster and without the use of cumbersome computations. An experimental verification of this approach was carried out using the example of a digital electrophotographic printing system.","PeriodicalId":297342,"journal":{"name":"2019 International Conference on Control, Artificial Intelligence, Robotics & Optimization (ICCAIRO)","volume":"56 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124462820","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 : 2019-05-01DOI: 10.1109/ICCAIRO47923.2019.00009
Shinya Kawabata, Jae Hoon Lee, S. Okamoto
A drone for inspecting aging infrastructures should have the capability to move near buildings where GPS signals cannot be received. In order to achieve a navigation technology not relying GPS, a system that can control the position of a drone in indoor environments was developed in this study. A position estimation algorithm using a tracking camera, Intel RealSense Tracking Camera T265, was employed to obtain the position information of the drone even in indoor environments without GPS information. Then, a system for controlling the position of the commanded target position was constructed and verified through experiments. In addition, a control algorithm to avoid obstacles by using horizontal movement was developed and tested with the developed system.
{"title":"Obstacle Avoidance Navigation Using Horizontal Movement for a Drone Flying in Indoor Environment","authors":"Shinya Kawabata, Jae Hoon Lee, S. Okamoto","doi":"10.1109/ICCAIRO47923.2019.00009","DOIUrl":"https://doi.org/10.1109/ICCAIRO47923.2019.00009","url":null,"abstract":"A drone for inspecting aging infrastructures should have the capability to move near buildings where GPS signals cannot be received. In order to achieve a navigation technology not relying GPS, a system that can control the position of a drone in indoor environments was developed in this study. A position estimation algorithm using a tracking camera, Intel RealSense Tracking Camera T265, was employed to obtain the position information of the drone even in indoor environments without GPS information. Then, a system for controlling the position of the commanded target position was constructed and verified through experiments. In addition, a control algorithm to avoid obstacles by using horizontal movement was developed and tested with the developed system.","PeriodicalId":297342,"journal":{"name":"2019 International Conference on Control, Artificial Intelligence, Robotics & Optimization (ICCAIRO)","volume":"23 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125943082","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 : 2019-05-01DOI: 10.1109/ICCAIRO47923.2019.00019
Dóra Varnyú, László Szirmay-Kalos
Positron emission tomography (PET) is a nuclear medicine imaging technique that is used to observe tissue metabolism by reconstructing the spatial distribution of the injected radioactive tracer. Due to constraints on the time and the radiation dose of the examination as well as limited scanner sensitivity, PET images usually suffer from a high level of noise. This paper focuses on the application of the guided filter for PET image denoising. After proposing several different guidance images, guided filter variants are compared with the median, the Gaussian and the bilateral filter in terms of image quality and speed. For dynamic PET reconstructions, a new approach, the parametric filtering is conceived, in which filtering is performed on the parameters of the kinetic model describing the radiotracer concentration. Finally, an efficient, guided-filter-based partial volume correction (PVC) method is proposed to restore accurate activity values that are blurred due to the partial volume effect.
{"title":"Improving Positron Emission Tomography with Guided Filtering","authors":"Dóra Varnyú, László Szirmay-Kalos","doi":"10.1109/ICCAIRO47923.2019.00019","DOIUrl":"https://doi.org/10.1109/ICCAIRO47923.2019.00019","url":null,"abstract":"Positron emission tomography (PET) is a nuclear medicine imaging technique that is used to observe tissue metabolism by reconstructing the spatial distribution of the injected radioactive tracer. Due to constraints on the time and the radiation dose of the examination as well as limited scanner sensitivity, PET images usually suffer from a high level of noise. This paper focuses on the application of the guided filter for PET image denoising. After proposing several different guidance images, guided filter variants are compared with the median, the Gaussian and the bilateral filter in terms of image quality and speed. For dynamic PET reconstructions, a new approach, the parametric filtering is conceived, in which filtering is performed on the parameters of the kinetic model describing the radiotracer concentration. Finally, an efficient, guided-filter-based partial volume correction (PVC) method is proposed to restore accurate activity values that are blurred due to the partial volume effect.","PeriodicalId":297342,"journal":{"name":"2019 International Conference on Control, Artificial Intelligence, Robotics & Optimization (ICCAIRO)","volume":"21 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127037845","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 : 2019-05-01DOI: 10.1109/ICCAIRO47923.2019.00011
K. Nohara, Jae Hoon Lee, S. Okamoto
A new drone for approaching and touching walls and ceilings to inspect aging buildings has been developed. In order to endow adhesion function to a drone, a specialized plate that exploits propeller's thrust force to generate adhering force near surface of ceiling and wall was designed. Its adhering force about three different design was investigated by using an experimental setup. The influences of propeller's rotating speed, plate's shape, and distance to surface were analyzed through experiments. Besides, the relation between plate design and power consumption was also studied. Furthermore, based on the experimental results, a drone with upper adhesion plate as well as side plate for adhering not only on the ceiling but also on the wall surfaces was developed. In addition, the practicability of the developed drone system was confirmed through experiments with manual operation.
{"title":"Experimental Evaluation of Adhesion Plate and Development of Novel Drone Capable of Adhering to Ceiling and Wall","authors":"K. Nohara, Jae Hoon Lee, S. Okamoto","doi":"10.1109/ICCAIRO47923.2019.00011","DOIUrl":"https://doi.org/10.1109/ICCAIRO47923.2019.00011","url":null,"abstract":"A new drone for approaching and touching walls and ceilings to inspect aging buildings has been developed. In order to endow adhesion function to a drone, a specialized plate that exploits propeller's thrust force to generate adhering force near surface of ceiling and wall was designed. Its adhering force about three different design was investigated by using an experimental setup. The influences of propeller's rotating speed, plate's shape, and distance to surface were analyzed through experiments. Besides, the relation between plate design and power consumption was also studied. Furthermore, based on the experimental results, a drone with upper adhesion plate as well as side plate for adhering not only on the ceiling but also on the wall surfaces was developed. In addition, the practicability of the developed drone system was confirmed through experiments with manual operation.","PeriodicalId":297342,"journal":{"name":"2019 International Conference on Control, Artificial Intelligence, Robotics & Optimization (ICCAIRO)","volume":"138 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124326713","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：