Search results for "Support Vector Machine"
showing 10 items of 306 documents
Deep learning for agricultural land use classification from Sentinel-2
2020
[ES] En el campo de la teledetección se ha producido recientemente un incremento del uso de técnicas de aprendizaje profundo (deep learning). Estos algoritmos se utilizan con éxito principalmente en la estimación de parámetros y en la clasificación de imágenes. Sin embargo, se han realizado pocos esfuerzos encaminados a su comprensión, lo que lleva a ejecutarlos como si fueran “cajas negras”. Este trabajo pretende evaluar el rendimiento y acercarnos al entendimiento de un algoritmo de aprendizaje profundo, basado en una red recurrente bidireccional de memoria corta a largo plazo (2-BiLSTM), a través de un ejemplo de clasificación de usos de suelo agrícola de la Comunidad Valenciana dentro d…
A Novel Time Series Kernel for Sequences Generated by LTI Systems
2017
The recent introduction of Hankelets to describe time series relies on the assumption that the time series has been generated by a vector autoregressive model (VAR) of order p. The success of Hankelet-based time series representations prevalently in nearest neighbor classifiers poses questions about if and how this representation can be used in kernel machines without the usual adoption of mid-level representations (such as codebook-based representations). It is also of interest to investigate how this representation relates to probabilistic approaches for time series modeling, and which characteristics of the VAR model a Hankelet can capture. This paper aims at filling these gaps by: deriv…
User Activity Recognition via Kinect in an Ambient Intelligence Scenario
2014
The availability of an ever-increasing kind of cheap, unobtrusive, sensing devices has stressed the need for new approaches to merge raw measurements in order to realize what is happening in the monitored environment. Ambient Intelligence (AmI) techniques exploit information about the environment state to adapt the environment itself to the users’ preferences. Even if traditional sensors allow a rough understanding of the users’ preferences, ad-hoc sensors are required to obtain a deeper comprehension of users’ habits and activities. In this paper we propose a framework to recognize users’ activities via a depth and RGB camera device, namely the Microsoft Kinect. The proposed approach takes…
Human Activity Recognition Process Using 3-D Posture Data
2015
In this paper, we present a method for recognizing human activities using information sensed by an RGB-D camera, namely the Microsoft Kinect. Our approach is based on the estimation of some relevant joints of the human body by means of the Kinect; three different machine learning techniques, i.e., K-means clustering, support vector machines, and hidden Markov models, are combined to detect the postures involved while performing an activity, to classify them, and to model each activity as a spatiotemporal evolution of known postures. Experiments were performed on Kinect Activity Recognition Dataset, a new dataset, and on CAD-60, a public dataset. Experimental results show that our solution o…
Super-resolution-based magnification of endothelium cells from biomicroscope videos of the cornea
2018
We present a practical, robust, and effective pipeline to compute a high-resolution (HR) image of the corneal endothelium starting from a low-resolution (LR) video sequence obtained with a general purpose slit lamp biomicroscope. An image quality typical of dedicated and more expensive confocal microscopes is achieved via software magnification by exploiting information redundancy in the video sequence. In particular, the HR image is generated from the best LR frames, obtained by identifying the most suitable endothelium video subsequence using a support vector machine-based learning approach, followed by a robust graph-based frame registration. Results on long, real sequences show that the…
Bayesian Network Based Classification of Mammography Structured Reports
2013
In modern medical domain, documents are created directly in electronic form and stored on huge databases containing documents, text in integral form and images. Retrieving right informations from these servers is challenging and, sometimes, this is very time consuming. Current medical technology do not provide a smart methodology classification of such documents based on their content. In this work the radiological structured reports are analysed classified and assigning an appropriate label. The text classifier is used to label a mammographic structured report. The experimental data are real clinical report coming from a hospital server. Analysing the structured report content, the classif…
A Multimodal People Recognition System for an Intelligent Environment
2011
In this paper, a multimodal system for recognizing people in intelligent environments is presented. Users are identified and tracked by detecting and recognizing voices and faces through cameras and microphones spread around the environment. This multimodal approach has been chosen to develop a flexible and cheap though reliable system, implemented through consumer electronics. Voice features are extracted through a short time spectrum analysis, while face features are extracted using the eigenfaces technique. The recognition task is achieved through the use of some Support Vector Machines, one per modality, that learn and classify the features of each person, while bindings between modalit…
A machine learning approach for user localization exploiting connectivity data
2016
The growing popularity of Location-Based Services (LBSs) has boosted research on cheaper and more pervasive localization systems, typically relying on such monitoring equipment as Wireless Sensor Networks (WSNs), which allow to re-use the same instrumentation both for monitoring and for localization without requiring lengthy off-line training. This work addresses the localization problem, exploiting knowledge acquired in sample environments, and extensible to areas not considered in advance. Localization is turned into a learning problem, solved by a statistical algorithm. Additionally, parameter tuning is fully automated thanks to its formulation as an optimization problem based only on co…
Text localization from photos
2009
In this paper a new text extraction algorithm is proposed. In real scenes the text is usually overlapped or is part of the background. To identify the text regions, in complex conditions, a method exploiting a “multi-resolution feature based method” for extracting text with undefined dimension has been developed. Once identified, the multi-resolution information are merged and skimmed through a set of Support Vector Machines (SVM). The tests and the comparisons with other techniques, performed on heterogeneous images, have shown the effectiveness of the proposed.
Automatic differentiation of melanoma from dysplastic nevi.
2015
International audience; Malignant melanoma causes the majority of deaths related to skin cancer. Nevertheless, it is the most treatable one, depending on its early diagnosis. The early prognosis is a challenging task for both clinicians and dermatologist, due to the characteristic similarities of melanoma with other skin lesions such as dysplastic nevi. In the past decades, several computerized lesion analysis algorithms have been proposed by the research community for detection of melanoma. These algorithms mostly focus on differentiating melanoma from benign lesions and few have considered the case of melanoma against dysplastic nevi. In this paper, we consider the most challenging task a…