This paper presents a new approach to the problem of semantic segmentation of digital images. We aim to improve the performance of some state-of-the-art approaches for the task. We exploit a new version of texton feature [28], which can encode image texture and object layout for learning a robust classifier. We propose to use a genetic algorithm for the learning parameters of weak classifiers in a boosting learning set up. We conducted extensive experiments on benchmark image datasets and compared the segmentation results with current proposed systems. The experimental results show that the performance of our system is comparable to, or even outperforms, those state-of-the-art algorithms. This is a promising approach as in this empirical study we used only texture-layout filter responses as feature and a basic setting of genetic algorithm. The framework is simple and can be extended and improved for many learning problems.
{"title":"Improving Image Segmentation Using Genetic Algorithm","authors":"Huynh Thi Thanh Binh, M. Loi, N. T. Thuy","doi":"10.1109/ICMLA.2012.134","DOIUrl":"https://doi.org/10.1109/ICMLA.2012.134","url":null,"abstract":"This paper presents a new approach to the problem of semantic segmentation of digital images. We aim to improve the performance of some state-of-the-art approaches for the task. We exploit a new version of texton feature [28], which can encode image texture and object layout for learning a robust classifier. We propose to use a genetic algorithm for the learning parameters of weak classifiers in a boosting learning set up. We conducted extensive experiments on benchmark image datasets and compared the segmentation results with current proposed systems. The experimental results show that the performance of our system is comparable to, or even outperforms, those state-of-the-art algorithms. This is a promising approach as in this empirical study we used only texture-layout filter responses as feature and a basic setting of genetic algorithm. The framework is simple and can be extended and improved for many learning problems.","PeriodicalId":157399,"journal":{"name":"2012 11th International Conference on Machine Learning and Applications","volume":"4 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2012-12-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125596193","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}
M. Kaden, D. Nebel, M. Riedel, Michael Biehl, T. Villmann
In the present paper we investigate the application of differentiable kernel for generalized matrix learning vector quantization as an alternative kernel-based classifier, which additionally provides classification dependent data visualization. We show that the concept of differentiable kernels allows a prototype description in the data space but equipped with the kernel metric. Moreover, using the visualization properties of the original matrix learning vector quantization we are able to optimize the class visualization by inherent visualization mapping learning also in this new kernel-metric data space.
{"title":"Differentiable Kernels in Generalized Matrix Learning Vector Quantization","authors":"M. Kaden, D. Nebel, M. Riedel, Michael Biehl, T. Villmann","doi":"10.1109/ICMLA.2012.231","DOIUrl":"https://doi.org/10.1109/ICMLA.2012.231","url":null,"abstract":"In the present paper we investigate the application of differentiable kernel for generalized matrix learning vector quantization as an alternative kernel-based classifier, which additionally provides classification dependent data visualization. We show that the concept of differentiable kernels allows a prototype description in the data space but equipped with the kernel metric. Moreover, using the visualization properties of the original matrix learning vector quantization we are able to optimize the class visualization by inherent visualization mapping learning also in this new kernel-metric data space.","PeriodicalId":157399,"journal":{"name":"2012 11th International Conference on Machine Learning and Applications","volume":"5 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2012-12-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123397217","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}
An error-driven adaptive model-based control system, for optimizing machine or assembly plant performance and operation under normal and fault conditions, is proposed. In such complex system it is imperative to differentiate between a system failure and a sensor failure or between process noise and measurement noise. In this paper, we present a comprehensive approach based on a hierarchical, multilevel control techniques. The approach is designed to provide sensor measurement validation, associates a degree of integrity with each measurement, identifies faulty sensors, and estimates the actual system states and sensor values in spite of faulty measurements. Using Virtual Machine Model concept, the method is achieved in three steps: state prediction, fault detection & sensor measurement and system online update or correction. A combination of flexible least square algorithm and adaptive Kalman filtering method are implemented to learn and predict system behavior. The experimental results show that the proposed model and algorithms can efficiently identify faulty components, reduce noise errors injected by sensors/system and thus providing self healing. The Virtual Machine Model (VMM) architecture described in this paper has proved to have several advantages over traditional models, the proposed model allows easy application provisioning, upgrades and maintenance, it provides fault tolerance, speedy disaster recovery and high availability platform.
{"title":"Error-Driven Adaptive, Virtual Machine Model-Based Control with High Availability Platform","authors":"Aman H. Bura, Bo Chen, Li Yu","doi":"10.1109/ICMLA.2012.133","DOIUrl":"https://doi.org/10.1109/ICMLA.2012.133","url":null,"abstract":"An error-driven adaptive model-based control system, for optimizing machine or assembly plant performance and operation under normal and fault conditions, is proposed. In such complex system it is imperative to differentiate between a system failure and a sensor failure or between process noise and measurement noise. In this paper, we present a comprehensive approach based on a hierarchical, multilevel control techniques. The approach is designed to provide sensor measurement validation, associates a degree of integrity with each measurement, identifies faulty sensors, and estimates the actual system states and sensor values in spite of faulty measurements. Using Virtual Machine Model concept, the method is achieved in three steps: state prediction, fault detection & sensor measurement and system online update or correction. A combination of flexible least square algorithm and adaptive Kalman filtering method are implemented to learn and predict system behavior. The experimental results show that the proposed model and algorithms can efficiently identify faulty components, reduce noise errors injected by sensors/system and thus providing self healing. The Virtual Machine Model (VMM) architecture described in this paper has proved to have several advantages over traditional models, the proposed model allows easy application provisioning, upgrades and maintenance, it provides fault tolerance, speedy disaster recovery and high availability platform.","PeriodicalId":157399,"journal":{"name":"2012 11th International Conference on Machine Learning and Applications","volume":"52 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2012-12-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126505217","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}
Álvaro Viloria, C. Chang, M. C. P. Socorro, J. Viloria
Landslides are processes of erosion of catastrophic character which alter the morphology of the landscape and affect people, productive land and infrastructure. Recently, there have been several attempts to apply neural networks to predict landscape susceptibility to landslides. However, the knowledge of the neural network is expressed in a mathematical model that does not allow establishing, intuitively, relationships between the factors causing landslides. This makes it difficult for experts to interpret the output of the network, to support their results with a set of inference rules. This limitation could be overcome by a model based on the FALCON neural network, which allows not only a classification for data clustering with fuzzy logic, but also generates a set of fuzzy rules from data training. For this reason, the FALCON-ART neural network has been implemented in this study to create a set of models of susceptibility to landslides on the watershed of the Caramacate River in north-central. The input data of the model included a landslide scar map from 1992, and variables derived from a digital elevation model and a SPOT-satellite image. A cross validation determined that the best result achieved a 74% success rate in predicting areas susceptible to landslides.
{"title":"Estimation of Susceptibility to Landslides Using Neural Networks Based on the FALCON-ART Model","authors":"Álvaro Viloria, C. Chang, M. C. P. Socorro, J. Viloria","doi":"10.1109/ICMLA.2012.122","DOIUrl":"https://doi.org/10.1109/ICMLA.2012.122","url":null,"abstract":"Landslides are processes of erosion of catastrophic character which alter the morphology of the landscape and affect people, productive land and infrastructure. Recently, there have been several attempts to apply neural networks to predict landscape susceptibility to landslides. However, the knowledge of the neural network is expressed in a mathematical model that does not allow establishing, intuitively, relationships between the factors causing landslides. This makes it difficult for experts to interpret the output of the network, to support their results with a set of inference rules. This limitation could be overcome by a model based on the FALCON neural network, which allows not only a classification for data clustering with fuzzy logic, but also generates a set of fuzzy rules from data training. For this reason, the FALCON-ART neural network has been implemented in this study to create a set of models of susceptibility to landslides on the watershed of the Caramacate River in north-central. The input data of the model included a landslide scar map from 1992, and variables derived from a digital elevation model and a SPOT-satellite image. A cross validation determined that the best result achieved a 74% success rate in predicting areas susceptible to landslides.","PeriodicalId":157399,"journal":{"name":"2012 11th International Conference on Machine Learning and Applications","volume":"30 8 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2012-12-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125700505","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}
Building prediction models for suggestive knowledge from multiple sources dynamically is of great interest from a clinical decision support point of view. This is valuable in situations where the local clinical data repository does not have sufficient number of records to draw conclusions from. However, due to privacy concerns, hospitals are reluctant to divulge patient records. Consequently, a distributed model building mechanism that can use just the statistics from multiple hospitals' databases is valuable. Our DIDT algorithm builds a model in that fashion. In this study, using National Inpatient Sample (NIS) data for 2009, we demonstrate that DIDT algorithm can be used to help collaboratively build a better decision-making model in situations where hospitals have small number of records that are insufficient to make good local models. Based on 262 attributes used for model building, we showed that 9 collaborating hospitals each with less than 100 cases of hospitalizations related to diabetes were able to achieve 9.9% improvement in accuracies of hospitalization prediction collectively using a distributed model as compared to relying on local models developed on their own. When relying on local risk prediction models for diabetes at these 9 hospitals, 159 of 357 patients were misclassified and prediction was impossible for another 16 patients. Our integrated model reduced the misclassification to 138 effectively providing accurate early diagnostics to 37 additional patients. We also introduce the concept of banding to improve DIDT algorithm so as to logically combine multiple hospitals when large number of hospitals is involved for reduction in cross-validation folds.
{"title":"Distributed Privacy Preserving Decision Support System for Predicting Hospitalization Risk in Hospitals with Insufficient Data","authors":"George Mathew, Zoran Obradovic","doi":"10.1109/ICMLA.2012.180","DOIUrl":"https://doi.org/10.1109/ICMLA.2012.180","url":null,"abstract":"Building prediction models for suggestive knowledge from multiple sources dynamically is of great interest from a clinical decision support point of view. This is valuable in situations where the local clinical data repository does not have sufficient number of records to draw conclusions from. However, due to privacy concerns, hospitals are reluctant to divulge patient records. Consequently, a distributed model building mechanism that can use just the statistics from multiple hospitals' databases is valuable. Our DIDT algorithm builds a model in that fashion. In this study, using National Inpatient Sample (NIS) data for 2009, we demonstrate that DIDT algorithm can be used to help collaboratively build a better decision-making model in situations where hospitals have small number of records that are insufficient to make good local models. Based on 262 attributes used for model building, we showed that 9 collaborating hospitals each with less than 100 cases of hospitalizations related to diabetes were able to achieve 9.9% improvement in accuracies of hospitalization prediction collectively using a distributed model as compared to relying on local models developed on their own. When relying on local risk prediction models for diabetes at these 9 hospitals, 159 of 357 patients were misclassified and prediction was impossible for another 16 patients. Our integrated model reduced the misclassification to 138 effectively providing accurate early diagnostics to 37 additional patients. We also introduce the concept of banding to improve DIDT algorithm so as to logically combine multiple hospitals when large number of hospitals is involved for reduction in cross-validation folds.","PeriodicalId":157399,"journal":{"name":"2012 11th International Conference on Machine Learning and Applications","volume":"19 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2012-12-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116148718","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}
From basic neuro-physiological evidences, it is now generally accepted that animals' walking control is subject to the combination function of central pattern generator(CPG) located at the spinal cords and reflexes from the peripheral stimulus. Since phase oscillators have the advantage of mathematical tractability, it's convenient to adjust the phase relationship between them. In this paper, coupled phase oscillators were designed to simulate CPG's behavior and establish vestibular reflex with feedbacks from accelerator sensors. Afterward, the synchronization condition of this proposed CPG model was studied. Forward and backward walking, gait transfers between trot and walk were realized as well. With feedbacks, AIBO detected uphill and downhill terrain and changed its posture automatically to fit for the new environment. Simulations were done in Webots to verify this method.
{"title":"CPG and Reflexes Combined Adaptive Walking Control for AIBO","authors":"Xianchao Zhao, Jiaqi Zhang, Chenkun Qi","doi":"10.1109/ICMLA.2012.81","DOIUrl":"https://doi.org/10.1109/ICMLA.2012.81","url":null,"abstract":"From basic neuro-physiological evidences, it is now generally accepted that animals' walking control is subject to the combination function of central pattern generator(CPG) located at the spinal cords and reflexes from the peripheral stimulus. Since phase oscillators have the advantage of mathematical tractability, it's convenient to adjust the phase relationship between them. In this paper, coupled phase oscillators were designed to simulate CPG's behavior and establish vestibular reflex with feedbacks from accelerator sensors. Afterward, the synchronization condition of this proposed CPG model was studied. Forward and backward walking, gait transfers between trot and walk were realized as well. With feedbacks, AIBO detected uphill and downhill terrain and changed its posture automatically to fit for the new environment. Simulations were done in Webots to verify this method.","PeriodicalId":157399,"journal":{"name":"2012 11th International Conference on Machine Learning and Applications","volume":"33 5","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2012-12-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"120994250","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}
As early as the late nineteenth century, scientists began research in author attribution, mostly by identifying the writing styles of authors. Following research over centuries has repeatedly demonstrated that people tend to have distinguishable writing styles. Today we not only have more authors, but we also have all different kinds of publications: journals, conferences, workshops, etc., covering different topics and requiring different writing formats. In spite of successful research in author attribution, no work has been carried out to find out whether publication venues are similarly distinguishable by their writing styles. Our work takes the first step into exploring this problem. By approaching the problem using a traditional classification method, we extract three types of writing style-based features and carry out detailed experiments in examining the different impacts among features, and classification techniques, as well as the influence of venue content, topics and genres. Experiments on real data from ACM and Cite Seer digital libraries demonstrate our approach to be an effective method in distinguishing venues in terms of their writing styles.
{"title":"Writing with Style: Venue Classification","authors":"Zaihan Yang, Brian D. Davison","doi":"10.1109/ICMLA.2012.50","DOIUrl":"https://doi.org/10.1109/ICMLA.2012.50","url":null,"abstract":"As early as the late nineteenth century, scientists began research in author attribution, mostly by identifying the writing styles of authors. Following research over centuries has repeatedly demonstrated that people tend to have distinguishable writing styles. Today we not only have more authors, but we also have all different kinds of publications: journals, conferences, workshops, etc., covering different topics and requiring different writing formats. In spite of successful research in author attribution, no work has been carried out to find out whether publication venues are similarly distinguishable by their writing styles. Our work takes the first step into exploring this problem. By approaching the problem using a traditional classification method, we extract three types of writing style-based features and carry out detailed experiments in examining the different impacts among features, and classification techniques, as well as the influence of venue content, topics and genres. Experiments on real data from ACM and Cite Seer digital libraries demonstrate our approach to be an effective method in distinguishing venues in terms of their writing styles.","PeriodicalId":157399,"journal":{"name":"2012 11th International Conference on Machine Learning and Applications","volume":"18 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2012-12-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125196817","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}
Class imbalance is prevalent in many real world datasets. It occurs when there are significantly fewer examples in one or more classes in a dataset compared to the number of instances in the remaining classes. When trained on highly imbalanced datasets, traditional machine learning techniques can often simply ignore the minority class(es) and label all instances as being of the majority class to maximize accuracy. This problem has been studied in many domains but there is little or no research related to the effect of class imbalance in fault data for condition monitoring of an ocean turbine. This study makes the first efforts in bridging that gap by providing insight into how class imbalance in vibration data can impact a learner's ability to reliably identify changes in the ocean turbine's operational state. To do so, we empirically evaluate the performances of three popular, but very different, machine learning algorithms when trained on four datasets with varying class distributions (one balanced and three imbalanced) to distinguish between a normal and an abnormal state. All data used in this study were collected from the testbed for an ocean turbine and were under sampled to simulate the different levels of imbalance. We find here, as in other domains, that the three learners seemed to suffer overall when trained on data with a highly skewed class distribution (with 0.1% examples in a faulty/abnormal state while the remaining 99.9% were captured in a normal operational state). It was noted, however, that the Logistic Regression and Decision Tree classifiers performed better when only 5% of the total number of examples were representative of an abnormal state (the remaining 95% therefore indicating normal operation) than they did when there was no imbalance present.
{"title":"Studying the Effect of Class Imbalance in Ocean Turbine Fault Data on Reliable State Detection","authors":"Janell Duhaney, T. Khoshgoftaar, Amri Napolitano","doi":"10.1109/ICMLA.2012.53","DOIUrl":"https://doi.org/10.1109/ICMLA.2012.53","url":null,"abstract":"Class imbalance is prevalent in many real world datasets. It occurs when there are significantly fewer examples in one or more classes in a dataset compared to the number of instances in the remaining classes. When trained on highly imbalanced datasets, traditional machine learning techniques can often simply ignore the minority class(es) and label all instances as being of the majority class to maximize accuracy. This problem has been studied in many domains but there is little or no research related to the effect of class imbalance in fault data for condition monitoring of an ocean turbine. This study makes the first efforts in bridging that gap by providing insight into how class imbalance in vibration data can impact a learner's ability to reliably identify changes in the ocean turbine's operational state. To do so, we empirically evaluate the performances of three popular, but very different, machine learning algorithms when trained on four datasets with varying class distributions (one balanced and three imbalanced) to distinguish between a normal and an abnormal state. All data used in this study were collected from the testbed for an ocean turbine and were under sampled to simulate the different levels of imbalance. We find here, as in other domains, that the three learners seemed to suffer overall when trained on data with a highly skewed class distribution (with 0.1% examples in a faulty/abnormal state while the remaining 99.9% were captured in a normal operational state). It was noted, however, that the Logistic Regression and Decision Tree classifiers performed better when only 5% of the total number of examples were representative of an abnormal state (the remaining 95% therefore indicating normal operation) than they did when there was no imbalance present.","PeriodicalId":157399,"journal":{"name":"2012 11th International Conference on Machine Learning and Applications","volume":"22 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2012-12-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125272469","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}
Convolutional neural network models have covered a broad scope of computer vision applications, achieving competitive performance with minimal domain knowledge. In this work, we apply such a model to a task designed to deter automated systems. We trained a convolutional neural network to distinguish between images of human faces from computer generated avatars as part of the ICMLA 2012 Face Recognition Challenge. The network achieved a classification accuracy of 99% on the Avatar CAPTCHA dataset. Furthermore, we demonstrated the potential of utilizing support vector machines on the same problem and achieved equally competitive performance.
{"title":"Convolutional Neural Networks Applied to Human Face Classification","authors":"Brian Cheung","doi":"10.1109/ICMLA.2012.177","DOIUrl":"https://doi.org/10.1109/ICMLA.2012.177","url":null,"abstract":"Convolutional neural network models have covered a broad scope of computer vision applications, achieving competitive performance with minimal domain knowledge. In this work, we apply such a model to a task designed to deter automated systems. We trained a convolutional neural network to distinguish between images of human faces from computer generated avatars as part of the ICMLA 2012 Face Recognition Challenge. The network achieved a classification accuracy of 99% on the Avatar CAPTCHA dataset. Furthermore, we demonstrated the potential of utilizing support vector machines on the same problem and achieved equally competitive performance.","PeriodicalId":157399,"journal":{"name":"2012 11th International Conference on Machine Learning and Applications","volume":"71 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2012-12-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127750725","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}
This paper addresses the challenging problem of scene classification in street-view georeferenced images of urban environments. More precisely, the goal of this task is semantic image classification, consisting in predicting in a given image, the presence or absence of a pre-defined class (e.g. shops, vegetation, etc.). The approach is based on the BOSSA representation, which enriches the Bag of Words (BoW) model, in conjunction with the Spatial Pyramid Matching scheme and kernel-based machine learning techniques. The proposed method handles problems that arise in large scale urban environments due to acquisition conditions (static and dynamic objects/pedestrians) combined with the continuous acquisition of data along the vehicle's direction, the varying light conditions and strong occlusions (due to the presence of trees, traffic signs, cars, etc.) giving rise to high intra-class variability. Experiments were conducted on a large dataset of high resolution images collected from two main avenues from the 12th district in Paris and the approach shows promising results.
{"title":"Classification of Urban Scenes from Geo-referenced Images in Urban Street-View Context","authors":"C. Iovan, David Picard, Nicolas Thome, M. Cord","doi":"10.1109/ICMLA.2012.171","DOIUrl":"https://doi.org/10.1109/ICMLA.2012.171","url":null,"abstract":"This paper addresses the challenging problem of scene classification in street-view georeferenced images of urban environments. More precisely, the goal of this task is semantic image classification, consisting in predicting in a given image, the presence or absence of a pre-defined class (e.g. shops, vegetation, etc.). The approach is based on the BOSSA representation, which enriches the Bag of Words (BoW) model, in conjunction with the Spatial Pyramid Matching scheme and kernel-based machine learning techniques. The proposed method handles problems that arise in large scale urban environments due to acquisition conditions (static and dynamic objects/pedestrians) combined with the continuous acquisition of data along the vehicle's direction, the varying light conditions and strong occlusions (due to the presence of trees, traffic signs, cars, etc.) giving rise to high intra-class variability. Experiments were conducted on a large dataset of high resolution images collected from two main avenues from the 12th district in Paris and the approach shows promising results.","PeriodicalId":157399,"journal":{"name":"2012 11th International Conference on Machine Learning and Applications","volume":"17 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2012-12-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128069475","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：