Fraunhofer-Institut für Produktionsanlagen und Konstruktionstechnik IPK

Filters

1 1
3 3
4 2
Reset filters

Settings
Now showing 1 - 6 of 6
  • Publication
    Rechnerunterstützte Optimierung - Möglichkeiten und Grenzen
    ( 2015)
    Menevidis, Zaharya
    ;
    Ajami, Mohamad
    Nach einer kurzen Einleitung über sogenannte lernende Systeme wird exemplarisch die rechnerunterstützte Optimierung am System InREAKT vorgestellt, dessen Architektur für die kontinuierliche Verbesserung der Gefahrendetektion und des Ereignismanagements verschiedene Lern- bzw. Optimierungsprinzipien vorsieht, die während der Phase der aktiven Detektion (online) wie auch außerhalb dieser Betriebsphase (offline) angewandt werden. Dafür werden unterschiedliche Werkzeuge und Verfahren bereitgestellt. Der ""Mensch"" ist dabei als ein integraler Bestandteil des Systems zu sehen, da es sich hier um ein maschinelles, überwachtes Weiterlernen mit Unterstützung eines ""Betreuers"" handelt. In der Online- Phase stehen diverse Dialogmöglichkeiten der Optimierungswerkzeuge im Ereignis-Management-System für die Bewertung und Korrektur der Detektion zur Verfügung. Danach warden Aktionen in dem Detektionslernsystem eingeleitet, die zur Aktualisierung der Methodendatenbank und Detektionsmodule dienen.
  • Publication
    Optische Sensorik - Intelligente Detektion gefährlicher Situationen
    ( 2014)
    Menevidis, Zaharya
    ;
    Ajami, Mohamad
    Vorgestellt werden die Komponenten und Verfahren eines verteilten intelligenten optischen Detektionssystems, das als Element eines generischen Gefahrenmanagementsystems in Fahrzeugen und Haltestellen vordefinierte interventionsbedürftige Szenen erkennt und meldet. Die optische Erfassung der Szene erfolgt über Tiefenbildsensoren, die Tiefensilhouetten und Gelenkdarstellungen generieren, sowie über RGB-Sensoren, welche die Bestimmung von farbbasierten charakteristischen Merkmalen und Individualisierung von Objekten gestatten. Daraus lässt sich die Deutung von Posen, Bewegungsmustern und Szenen ableiten. Die Intelligenz liegt in den Verfahren der Posenbeschreibung (Motion Description Language MDL, Neuronale Netze) und Zustandsinterpretationen (Scene Activity Modelling SAM) mit den Möglichkeiten der Integration von Lernverfahren in den Online- (Auswertung und Bewertung während der Detektion) sowie Offline-Phasen (Veränderung der Zustandsinterpretationen).
  • Publication
    Automatic detection of situations for intervention using 3D image reconstruction and template matching
    ( 2014)
    Ajami, Mohamad
    ;
    Menevidis, Zaharya
    ;
    Gerson, Christian
    Due to the public transportation companies' need to ensure the safety and security of the passengers, the recent years have witnessed a substantial increase in the number of surveillance cameras installed in these stations. Regarding to this increasing number, the companies automatically face the problem that human operators have some difficulties thoroughly monitoring all cameras' outputs meticulously. In this paper, we will present a system that will monitor a selected zone in a public station and will serve solely as an assisting component for the surveillance system operator who will take the final appropriate measures for intervention. The system will be analysing the scenes in real time by reconstructing the foreground objects in 3D space and using a series of image processing methods to detect a predefined ""Situation in Need of Intervention"" (SNI). This paper will describe three main SNIs (falling/lying on the ground, act of human aggression and unattended objects) and will show the methodology that was used to classify a scene under these three SNIs.
  • Publication
    D 4.1: Observatory descriptive report
    (Fraunhofer IPK, 2014)
    Menevidis, Zaharya
    ;
    Hahne, Michael
    ;
    Ajami, Mohamad
    ;
    Fairweather, Ben
    ;
    Smagas, Kostas
    ;
    Giambene, Giovanni
    ;
    Hahne, Michael
    ;
    Menevidis, Zaharya
    Deliverable 4.1 provides the specification for the ""Observatory for International Responsible Research and Innovation Coordination"". Its purpose is to enable the implementation of the Observatory in task 4.2. Therefore the technical as well as procedural requirements and prerequisites for the Observatory have been defined in detail. As the Observatory is intended to harness the involvement of the broader network of researchers and innovators, their participation in the design of it should maximise the chances of it being a tool they take ownership of. Therefore the gathering of requirements was not only limited to the description of work but included the expertise of the participants of the Responsibility project and was extended by the integration of feedback from the other current EU RRI project members. Chapter 2 of this deliverable provides an overview of the requirements incorporated in the specification process. As it is good practice in software design use cases have been developed based on these requirements. These have been described in chapter 3. The purpose of use cases is to specify interaction processes, identify variation and failure scenarios as well as technical functionalities and procedural modalities that need an in depth specification. These in depth specifications have been described in chapter 4. They must comprise the input data needed as well as a definition of their output and to where that data will be passed on. Apart from that it contains the specification of the processing of that data, of the interface design and of the information necessary to explain the functionality to the user.
  • Publication
    Verbundprojekt: ADIS
    (Fraunhofer IPK, 2014)
    Menevidis, Zaharya
    ;
    Ajami, Mohamad
  • Publication
    Vision based fire detection using colour variance
    ( 2013)
    Ajami, Mohamad
    ;
    Menevidis, Zaharya
    In recent years, there have been a variety of attempts in research and industry to solve various problems by using the non-contact detection and interpretation possibilities of computer vision, as one image offers the floor to analyse different causes and sources in parallel. It is logical to apply an integrated approach via image and video not only to venture into new and innovative fields, but also to optimize and improve existing solutions with other sensors and to solve contemporary problems. In this paper, we will deal with the integrated approach for optimizing and improving fire detection. The detection path is going to be solely based on the colour features of household fires in addition to a colour independent feature of fire that owes to enhance the detection result. The end goal is to achieve real time fire detection and obtain a low false positive rate.