SIM, a Semantic Instrumentation and Monitoring solution for large scale reasoning systems
One central task to the idea of Semantic Web is reasoning over semantic descriptions of web pages... more One central task to the idea of Semantic Web is reasoning over semantic descriptions of web pages and information items available on the Web. A flagship project that is advancing the state of the art in reasoning with Web-scale data is the Large Knowledge Collider (LarKC). Having a pluggable architecture, LarKC enables the interested users to test their reasoning approaches with very little overhead. In this context, instrumenting and monitoring of the large scale reasoning systems and their components becomes essential for verifying and assuring high performance, adaptability and well-functioning. These aspects are in the end vital for any reasoning experiment. We introduce SIM, Semantic Instrumentation and Monitoring, a semantic-based instrumentation and monitoring solution. SIM enables the instrumentation and monitoring of LarKC applications in particular and any large scale reasoning system in general. It offers the means for developers to specify the metrics of interest, to instrument the code, to collect and observe how well the system and its components are performing. We identify a large set of relevant metrics for monitoring and provide ontological models for them. Finally we discuss the architecture and the role of each component and tool which is part of SIM.
Detecting the parts of a vehicle represents a topic of major interest for computer vision applica... more Detecting the parts of a vehicle represents a topic of major interest for computer vision applications, especially for precrash systems. This paper proposes an artificial vision based technique that identifies the pillars of the lateral viewed cars. The novelty of the approach resides in the multi-layer classification scheme applied within the context of a stereo-based object detection system. From all the objects deetected by stereovision the side viewed cars are recognized, and for them the pillars are identified. This process of pillar identification is the result of a multi-layer classification that comprises: a rough object hypothesis refinement that selects only those objects that are likely to have one or two wheels, followed by an adaptive boosting classifier build using histograms of oriented gradient features. The boosted classifier realizes a fine selection of the wheel-based hypotheses and discriminates between side viewed vehicles and other objects in a traffic scene. The last step consists in the construction of a geometrical model of the pillars' region of interest for the identified side vehicles.
A comparative study of pedestrian detection methods using classical Haar and HoG features versus bag of words model computed from Haar and HoG features
The bag of words model has been actively adopted by content based image retrieval and image annot... more The bag of words model has been actively adopted by content based image retrieval and image annotation techniques. We employ this model for the particular task of pedestrian detection in two dimensional images, producing this way a novel approach to pedestrian detection. The experiments we have done in this paper compare the behavior of discriminative recognition approaches that use AdaBoost on codebook features versus Adaboost trained on primitive features that may be extracted from a two dimensional image. By primitive features we refer in this paper to Haar features and Histogram of Oriented Gradients both being extremely used in object recognition in general and in pedestrian detection in particular. The conclusion of our experiments is that the codebook representation performs better than the primitive feature representation.
Reasoning is central to the idea of the Semantic Web and ontologies, however, the fundamental pri... more Reasoning is central to the idea of the Semantic Web and ontologies, however, the fundamental principles of reasoning -soundness and completenessdo not match the reality of the (Semantic) Web that is ruled by contradicting and incomplete data and claims. Furthermore, logical reasoning is strong for rather small numbers of axioms and facts, while the Web is growing at an impressive speed and hence offering tremendous amounts of data. 'Reasearch' is a novel idea of combining reasoning with information retrieval methods (search) in order to respond to the requirements and constraints implied by reasoning on the Semantic Web. The Large Knowledge Collider is a modular reasoning platform that allows for reasearching with Web-scale data, and instrumenting and monitoring the platform is essential for verifying and assuring high performance, adaptability and wellfunctioning. These aspects are vital for experiments running on the platform.
In this paper we introduce a system for semantic understanding of traffic scenes. The system dete... more In this paper we introduce a system for semantic understanding of traffic scenes. The system detects objects in video images captured in real vehicular traffic situations, classifies them, maps them to the OpenCyc 1 ontology and finally generates descriptions of the traffic scene in CycL or cvasinatural language. We employ meta-classification methods based on AdaBoost and Random forest algorithms for identifying interest objects like: cars, pedestrians, poles in traffic and we derive a set of annotations for each traffic scene. These annotations are mapped to OpenCyc concepts and predicates, spatiotemporal rules for object classification and scene understanding are then asserted in the knowledge base. Finally, we show that the system performs well in understanding traffic scene situations and summarizing them. The novelty of the approach resides in the combination of stereo-based object detection and recognition methods with logic based spatio-temporal reasoning.
Modeling the performance of large scale systems is the core idea of this paper.We focus on modeli... more Modeling the performance of large scale systems is the core idea of this paper.We focus on modeling the performance specific behavior of LarKC1- The Large Knowledge Collider a platform for large scale integrated reasoning and Web-search. A set of instrumentation and monitoring tools are employed to collect metrics related to execution time, resources, and specific platform measurements like running workflows and plug-ins. Our method performs machine learning on top of instrumented data and tries to find relations between input defined metrics and output metrics that describe the instrumentation observations of the LarKC platform, plug-ins or workflows. The proposed method is a combination of clustering and regression techniques.
Object recognition from images is one of the essential problems in automatic image processing. In... more Object recognition from images is one of the essential problems in automatic image processing. In this paper we focus specifically on nearest neighbor methods, which are widely used in many practical applications, not necessarily related to image data. It has recently come to attention that high dimensional data also exhibit high hubness, which essentially means that some very influential data points appear and these points are referred to as hubs. Unsurprisingly, hubs play a very important role in the nearest neighbor classification. We examine the hubness of various image data sets, under several different feature representations. We also show that it is possible to exploit the observed hubness and improve the recognition accuracy.
SIM, a Semantic Instrumentation and Monitoring solution for large scale reasoning systems
One central task to the idea of Semantic Web is reasoning over semantic descriptions of web pages... more One central task to the idea of Semantic Web is reasoning over semantic descriptions of web pages and information items available on the Web. A flagship project that is advancing the state of the art in reasoning with Web-scale data is the Large Knowledge Collider (LarKC). Having a pluggable architecture, LarKC enables the interested users to test their reasoning approaches with very little overhead. In this context, instrumenting and monitoring of the large scale reasoning systems and their components becomes essential for verifying and assuring high performance, adaptability and well-functioning. These aspects are in the end vital for any reasoning experiment. We introduce SIM, Semantic Instrumentation and Monitoring, a semantic-based instrumentation and monitoring solution. SIM enables the instrumentation and monitoring of LarKC applications in particular and any large scale reasoning system in general. It offers the means for developers to specify the metrics of interest, to instrument the code, to collect and observe how well the system and its components are performing. We identify a large set of relevant metrics for monitoring and provide ontological models for them. Finally we discuss the architecture and the role of each component and tool which is part of SIM.
Detecting the parts of a vehicle represents a topic of major interest for computer vision applica... more Detecting the parts of a vehicle represents a topic of major interest for computer vision applications, especially for precrash systems. This paper proposes an artificial vision based technique that identifies the pillars of the lateral viewed cars. The novelty of the approach resides in the multi-layer classification scheme applied within the context of a stereo-based object detection system. From all the objects deetected by stereovision the side viewed cars are recognized, and for them the pillars are identified. This process of pillar identification is the result of a multi-layer classification that comprises: a rough object hypothesis refinement that selects only those objects that are likely to have one or two wheels, followed by an adaptive boosting classifier build using histograms of oriented gradient features. The boosted classifier realizes a fine selection of the wheel-based hypotheses and discriminates between side viewed vehicles and other objects in a traffic scene. The last step consists in the construction of a geometrical model of the pillars' region of interest for the identified side vehicles.
A comparative study of pedestrian detection methods using classical Haar and HoG features versus bag of words model computed from Haar and HoG features
The bag of words model has been actively adopted by content based image retrieval and image annot... more The bag of words model has been actively adopted by content based image retrieval and image annotation techniques. We employ this model for the particular task of pedestrian detection in two dimensional images, producing this way a novel approach to pedestrian detection. The experiments we have done in this paper compare the behavior of discriminative recognition approaches that use AdaBoost on codebook features versus Adaboost trained on primitive features that may be extracted from a two dimensional image. By primitive features we refer in this paper to Haar features and Histogram of Oriented Gradients both being extremely used in object recognition in general and in pedestrian detection in particular. The conclusion of our experiments is that the codebook representation performs better than the primitive feature representation.
Reasoning is central to the idea of the Semantic Web and ontologies, however, the fundamental pri... more Reasoning is central to the idea of the Semantic Web and ontologies, however, the fundamental principles of reasoning -soundness and completenessdo not match the reality of the (Semantic) Web that is ruled by contradicting and incomplete data and claims. Furthermore, logical reasoning is strong for rather small numbers of axioms and facts, while the Web is growing at an impressive speed and hence offering tremendous amounts of data. 'Reasearch' is a novel idea of combining reasoning with information retrieval methods (search) in order to respond to the requirements and constraints implied by reasoning on the Semantic Web. The Large Knowledge Collider is a modular reasoning platform that allows for reasearching with Web-scale data, and instrumenting and monitoring the platform is essential for verifying and assuring high performance, adaptability and wellfunctioning. These aspects are vital for experiments running on the platform.
In this paper we introduce a system for semantic understanding of traffic scenes. The system dete... more In this paper we introduce a system for semantic understanding of traffic scenes. The system detects objects in video images captured in real vehicular traffic situations, classifies them, maps them to the OpenCyc 1 ontology and finally generates descriptions of the traffic scene in CycL or cvasinatural language. We employ meta-classification methods based on AdaBoost and Random forest algorithms for identifying interest objects like: cars, pedestrians, poles in traffic and we derive a set of annotations for each traffic scene. These annotations are mapped to OpenCyc concepts and predicates, spatiotemporal rules for object classification and scene understanding are then asserted in the knowledge base. Finally, we show that the system performs well in understanding traffic scene situations and summarizing them. The novelty of the approach resides in the combination of stereo-based object detection and recognition methods with logic based spatio-temporal reasoning.
Modeling the performance of large scale systems is the core idea of this paper.We focus on modeli... more Modeling the performance of large scale systems is the core idea of this paper.We focus on modeling the performance specific behavior of LarKC1- The Large Knowledge Collider a platform for large scale integrated reasoning and Web-search. A set of instrumentation and monitoring tools are employed to collect metrics related to execution time, resources, and specific platform measurements like running workflows and plug-ins. Our method performs machine learning on top of instrumented data and tries to find relations between input defined metrics and output metrics that describe the instrumentation observations of the LarKC platform, plug-ins or workflows. The proposed method is a combination of clustering and regression techniques.
Object recognition from images is one of the essential problems in automatic image processing. In... more Object recognition from images is one of the essential problems in automatic image processing. In this paper we focus specifically on nearest neighbor methods, which are widely used in many practical applications, not necessarily related to image data. It has recently come to attention that high dimensional data also exhibit high hubness, which essentially means that some very influential data points appear and these points are referred to as hubs. Unsurprisingly, hubs play a very important role in the nearest neighbor classification. We examine the hubness of various image data sets, under several different feature representations. We also show that it is possible to exploit the observed hubness and improve the recognition accuracy.
Uploads
Papers by Raluca Brehar