Browsing by Author "Destefanis, Eduardo"

Filter results by typing the first few letters
Now showing 1 - 4 of 4
  • Thumbnail Image
    Item
    Determining diameter of animal textile fiber using image processing techniques
    (2014) Arcidiácono, Marcelo J.M.; Destefanis, Eduardo; Constable, Leticia E.; Vázquez, Juan Carlos
    In the context of productive sustainability of animal textile fibers, the possibility to achieve safe and inexpensive methods allowing to measure fiber quality is paramount for farmers pretending to compete in local and international markets. Using an image of a longitudinal cut of camelid fibers, we propose a computational method in which a Hierarchical Temporary Memory algorithm recognizes and identifies each of these fibers and a modified Gabor filter enhances and reconstructs the input image in order to reduce measurement errors of the mean diameter due to crossovers and out of focus objects. In this paper this procedure is described and results are discussed. The obtained results agree with the laboratory measurements.
  • Thumbnail Image
    Item
    Implementation and performance evaluation of UKF for Simultaneous Localization and Mapping
    (Universidad Nacional del Centro de la Provincia de Buenos Aires, 2012) Pérez Paina, Gonzalo; Paz, Claudio; Baudino, Martín; Delfino, Ariel; Destefanis, Eduardo
    The implementation of the Unscented Kalman Filter (UKF) for SLAM estimation is described. The UKF presents comparatively lower linearization error with respect to the typically used Extended Kalman Filter (EKF). The algorithm described in detail implements an UKFSLAM to build a feature-based map representation, for a mobile robot using a laser rangefinder. An evaluation comparing the UKF/EKF-SLAM estimations is shown. Results demonstrate a better performance of the UKF in terms of both robot pose and map feature positions estimation, besides the fact that UKF is easier to implement than EKF
  • Thumbnail Image
    Item
    Red neuronal multiescala para clasificación de la calidad vocal
    (Universidad Tecnológica Nacional Regional Córdoba., 2021) García , Mario Alejandro; Rosset, Ana Lorena; Destefanis, Eduardo
    La valoración de la calidad vocal mediante el análisis audio-perceptual es parte de la rutina clínica de evaluación de pacientes con trastornos de la voz. La debilidad de este método reside en la subjetividad y en la necesidad de que sea realizada por oyentes experimentados. Este proyecto tiene como objetivo la realización de una clasificación automática de la calidad vocal, valuada en la escala GRBAS, mediante la aplicación de técnicas de aprendizaje profundo sobre voces grabadas. Particularmente, en este trabajo se muestran los resultados del diseño de una red neuronal multiescala para la clasificación de la calidad vocal.
  • Thumbnail Image
    Item
    Trainable windowing coefficients in DNN for raw audio classification
    (Cloud computing, big data y emerging topics, 2020) García , Mario Alejandro; Destefanis, Eduardo; Rosset, Ana Lorena
    An artificial neural network for audio classification is pro posed. This includes the windowing operation of raw audio and the calculation of the power spectrogram. A windowing layer is initialized with a hann window and its weights are adapted during training. The non-trainable weights of spectrogram calculation are initialized with the discrete Fourier transform coefficients. The tests are performed on the Speech Commands dataset. Results show that adapting the windowing coefficients produces a moderate accuracy improvement. It is concluded that the gradient of the error function can be propagated through the neural calculation of the power spectrum. It is also concluded that the training of the windowing layer improves the model’s ability to general iz