Logo des Repositoriums

Auflistung nach Schlagwort "image recognition"

1 - 5 von 5
  • Konferenzbeitrag
    The application of image recognition methods to improve the performance of waste-to-energy plantsplants
    (EnviroInfo 2022, 2022) Schwark, Fenja; Garmatter, Henriette; Davila, Maria; Dawel, Lisa; Pehlken, Alexandra; Cyris, Fabian; Scharf, Roland
    In this paper, we present an image recognition method to improve the performance of waste-to-energy plants. Thermal treatment of waste in waste-to-energy plants is central for the treatment of municipal solid waste. The heterogeneous nature of municipal solid waste results in a fluctuating lower calorific value to which plant operation must be adapted. Compensating for drastic changes in the lower calorific value is challenging for plant operation and can require short-term interventions. Estimating the lower calorific value prior to the combustion process should reduce the number of short-term interventions. In this work, we propose a process-engineering approach to estimate the lower calorific value of waste as a new application of image recognition in waste-to-energy plants. The method is implemented using videos and sensor data from a case study in a real waste-to-energy plant in Germany.
  • Konferenzbeitrag
    Detection of snow-coverage on PV-modules with images based on CNN-techniques
    (EnviroInfo 2022, 2022) Hepp, Dennis; Hempelmann, Sebastian; Behrens, Grit; Friedrich, Werner
    The transition from fossil fuels to renewable energy is considered as very meaningful to mitigate climate change. To integrate weather-dependent energies firmly into the power grid, a forecast of the energy yield is very important. This paper is about renewable energy generation by photovoltaic (PV) systems. The yield of PV-systems depends not only on weather conditions, but in wintertime also on the additional factor “snow cover”. The aim of this work is to detect snow cover on photovoltaic plants to support the energy yield forecast. For this purpose, images of a PV-plant with and without snow cover are used for feature extraction and then analyzed by using a convolutional neural network (CNN).
  • Konferenzbeitrag
    Doorfinder oder "der digitale Blindenhund aus der Hosentasche"
    (Mensch und Computer 2019 - Usability Professionals, 2019) Sümnick, Marcus
    Die Orientierung von Menschen mit Sehbehinderung im öffentlichen Nähverkehr stellt noch immer eine Herausforderung dar. Vor allem wenn sie nicht durch Assistenzhunde unterstützt werden können. Unser Ansatz zeigt mit dem "digitalen Blindenhund" auf Potential für einen alternativen Ansatz auf. Durch das Zusammenbringen von leistungsstarken Smartphones und modernen Algorithmen wird eine Unterstützung ermöglicht, die vor wenigen Jahren nicht denkbar war. Das konkrete Anwendungsbeispiel "Doorfinder" stellen prototypisch den Lösungsvorschlag für das beschriebene Problem vieler Menschen mit Sehbehinderungen dar. Durch unseren den UX-Ansatz, der auf den Nutzungskontext fokussierten ist, ermöglicht es der "Doorfinder" blinden Menschen ohne fremde Unterstützung den Einstieg in Züge zu finden. Wir sehen unzählige weitere Möglichkeiten im öffentlichen und privaten Sektor, in der unser Ansatz zum Einsatz kommen kann. Lassen sie uns gemeinsam bewerten, wo wir starten können.
  • Konferenzbeitrag
    Fake war crime image detection by reverse image search
    (Datenbanksysteme für Business, Technologie und Web (BTW 2017) - Workshopband, 2017) Askinadze, Alexander
    In the media, images of war crimes are often shared, which in reality come from other contexts or other war sites. In this paper an approach is proposed to detect duplicate or fake war crime images. For this, the bag of visual words model is used in conjunction with localized soft assignment coding and the k-nn classifier. For evaluation, a data set with 600 images of war crimes was crawled. Different distances and parameters were used for evaluation. Unmodified images can be recognized with this approach with 100% accuracy. Rotated and scaled images can also be detected with nearly 100% accuracy. Modifications like cropping or the combination of scaling and cropping ensure significantly smaller accuracy results. The run time was investigated and it was found that about 3000 images per second can be processed on an Intel Core i5 processor.
  • Workshopbeitrag
    Using hash visualization for real-time user-governed password validation
    (Mensch und Computer 2019 - Workshopband, 2019) Fietkau, Julian; Balthasar, Mandy
    Building upon work by Perrig & Song [21], we propose a novel hash visualization algorithm and examine its usefulness for user-governed password validation in real time. In contrast to network-based password authentication and the best practices for security which have been developed with that paradigm in mind, we are concerned with use cases that require user-governed password validation in nonnetworked untrusted contexts, i.e. to allow a user to verify that they have typed their password correctly without ever storing a record of the correct password between sessions (not even a hash). To that end, we showcase a newly designed hash visualization algorithm named MosaicVisualHash and describe how hash visualization algorithms can be used to perform user-governed password validation. We also provide a set of design recommendations for systems where hash visualization for password validation is performed in real time, i.e. as the user is in the process of typing their password.