Gender classification in large databases

[EN]In this paper, we address the challenge of gender classi - cation using large databases of images with two goals. The rst objective is to evaluate whether the error rate decreases compared to smaller databases. The second goal is to determine if the classi er that provides the best classi cation rate for one database, improves the classi cation results for other databases, that is, the cross-database performance.

Road environment modeling using robust perspective analysis and recursive Bayesian segmentation

Recently, vision-based advanced driver-assistance systems (ADAS) have received a new increased interest to enhance driving safety. In particular, due to its high performance–cost ratio, mono-camera systems are arising as the main focus of this field of work. In this paper we present a novel on-board road modeling and vehicle detection system, which is a part of the result of the European I-WAY project. The system relies on a robust estimation of the perspective of the scene, which adapts to the dynamics of the vehicle and generates a stabilized rectified image of the road plane. This rectified plane is used by a recursive Bayesian classi- fier, which classifies pixels as belonging to different classes corresponding to the elements of interest of the scenario. This stage works as an intermediate layer that isolates subsequent modules since it absorbs the inherent variability of the scene. The system has been tested on-road, in different scenarios, including varied illumination and adverse weather conditions, and the results have been proved to be remarkable even for such complex scenarios.

Vision-based vehicle detection and tracking with a mobile camera using a statistical framework

En esta tesis se aborda la detección y el seguimiento automático de vehículos mediante técnicas de visión artificial con una cámara monocular embarcada. Este problema ha suscitado un gran interés por parte de la industria automovilística y de la comunidad científica ya que supone el primer paso en aras de la ayuda a la conducción, la prevención de accidentes y, en última instancia, la conducción automática. A pesar de que se le ha dedicado mucho esfuerzo en los últimos años, de momento no se ha encontrado ninguna solución completamente satisfactoria y por lo tanto continúa siendo un tema de investigación abierto. Los principales problemas que plantean la detección y seguimiento mediante visión artificial son la gran variabilidad entre vehículos, un fondo que cambia dinámicamente debido al movimiento de la cámara, y la necesidad de operar en tiempo real. En este contexto, esta tesis propone un marco unificado para la detección y seguimiento de vehículos que afronta los problemas descritos mediante un enfoque estadístico. El marco se compone de tres grandes bloques, i.e., generación de hipótesis, verificación de hipótesis, y seguimiento de vehículos, que se llevan a cabo de manera secuencial. No obstante, se potencia el intercambio de información entre los diferentes bloques con objeto de obtener el máximo grado posible de adaptación a cambios en el entorno y de reducir el coste computacional. Para abordar la primera tarea de generación de hipótesis, se proponen dos métodos complementarios basados respectivamente en el análisis de la apariencia y la geometría de la escena. Para ello resulta especialmente interesante el uso de un dominio transformado en el que se elimina la perspectiva de la imagen original, puesto que este dominio permite una búsqueda rápida dentro de la imagen y por tanto una generación eficiente de hipótesis de localización de los vehículos. Los candidatos finales se obtienen por medio de un marco colaborativo entre el dominio original y el dominio transformado. Para la verificación de hipótesis se adopta un método de aprendizaje supervisado. Así, se evalúan algunos de los métodos de extracción de características más populares y se proponen nuevos descriptores con arreglo al conocimiento de la apariencia de los vehículos. Para evaluar la efectividad en la tarea de clasificación de estos descriptores, y dado que no existen bases de datos públicas que se adapten al problema descrito, se ha generado una nueva base de datos sobre la que se han realizado pruebas masivas. Finalmente, se presenta una metodología para la fusión de los diferentes clasificadores y se plantea una discusión sobre las combinaciones que ofrecen los mejores resultados. El núcleo del marco propuesto está constituido por un método Bayesiano de seguimiento basado en filtros de partículas. Se plantean contribuciones en los tres elementos fundamentales de estos filtros: el algoritmo de inferencia, el modelo dinámico y el modelo de observación. En concreto, se propone el uso de un método de muestreo basado en MCMC que evita el elevado coste computacional de los filtros de partículas tradicionales y por consiguiente permite que el modelado conjunto de múltiples vehículos sea computacionalmente viable. Por otra parte, el dominio transformado mencionado anteriormente permite la definición de un modelo dinámico de velocidad constante ya que se preserva el movimiento suave de los vehículos en autopistas. Por último, se propone un modelo de observación que integra diferentes características. En particular, además de la apariencia de los vehículos, el modelo tiene en cuenta también toda la información recibida de los bloques de procesamiento previos. El método propuesto se ejecuta en tiempo real en un ordenador de propósito general y da unos resultados sobresalientes en comparación con los métodos tradicionales. ABSTRACT This thesis addresses on-road vehicle detection and tracking with a monocular vision system. This problem has attracted the attention of the automotive industry and the research community as it is the first step for driver assistance and collision avoidance systems and for eventual autonomous driving. Although many effort has been devoted to address it in recent years, no satisfactory solution has yet been devised and thus it is an active research issue. The main challenges for vision-based vehicle detection and tracking are the high variability among vehicles, the dynamically changing background due to camera motion and the real-time processing requirement. In this thesis, a unified approach using statistical methods is presented for vehicle detection and tracking that tackles these issues. The approach is divided into three primary tasks, i.e., vehicle hypothesis generation, hypothesis verification, and vehicle tracking, which are performed sequentially. Nevertheless, the exchange of information between processing blocks is fostered so that the maximum degree of adaptation to changes in the environment can be achieved and the computational cost is alleviated. Two complementary strategies are proposed to address the first task, i.e., hypothesis generation, based respectively on appearance and geometry analysis. To this end, the use of a rectified domain in which the perspective is removed from the original image is especially interesting, as it allows for fast image scanning and coarse hypothesis generation. The final vehicle candidates are produced using a collaborative framework between the original and the rectified domains. A supervised classification strategy is adopted for the verification of the hypothesized vehicle locations. In particular, state-of-the-art methods for feature extraction are evaluated and new descriptors are proposed by exploiting the knowledge on vehicle appearance. Due to the lack of appropriate public databases, a new database is generated and the classification performance of the descriptors is extensively tested on it. Finally, a methodology for the fusion of the different classifiers is presented and the best combinations are discussed. The core of the proposed approach is a Bayesian tracking framework using particle filters. Contributions are made on its three key elements: the inference algorithm, the dynamic model and the observation model. In particular, the use of a Markov chain Monte Carlo method is proposed for sampling, which circumvents the exponential complexity increase of traditional particle filters thus making joint multiple vehicle tracking affordable. On the other hand, the aforementioned rectified domain allows for the definition of a constant-velocity dynamic model since it preserves the smooth motion of vehicles in highways. Finally, a multiple-cue observation model is proposed that not only accounts for vehicle appearance but also integrates the available information from the analysis in the previous blocks. The proposed approach is proven to run near real-time in a general purpose PC and to deliver outstanding results compared to traditional methods.

Comparasion of road traffic emission models in Madrid (Spain)

Many cities in Europe have difficulties to meet the air quality standards set by the European legislation, most particularly the annual mean Limit Value for NO2. Road transport is often the main source of air pollution in urban areas and therefore, there is an increasing need to estimate current and future traffic emissions as accurately as possible. As a consequence, a number of specific emission models and emission factors databases have been developed recently. They present important methodological differences and may result in largely diverging emission figures and thus may lead to alternative policy recommendations. This study compares two approaches to estimate road traffic emissions in Madrid (Spain): the COmputer Programme to calculate Emissions from Road Transport (COPERT4 v.8.1) and the Handbook Emission Factors for Road Transport (HBEFA v.3.1), representative of the ‘average-speed’ and ‘traffic situation’ model types respectively. The input information (e.g. fleet composition, vehicle kilometres travelled, traffic intensity, road type, etc.) was provided by the traffic model developed by the Madrid City Council along with observations from field campaigns. Hourly emissions were computed for nearly 15 000 road segments distributed in 9 management areas covering the Madrid city and surroundings. Total annual NOX emissions predicted by HBEFA were a 21% higher than those of COPERT. The discrepancies for NO2 were lower (13%) since resulting average NO2/NOX ratios are lower for HBEFA. The larger differences are related to diesel vehicle emissions under “stop & go” traffic conditions, very common in distributor/secondary roads of the Madrid metropolitan area. In order to understand the representativeness of these results, the resulting emissions were integrated in an urban scale inventory used to drive mesoscale air quality simulations with the Community Multiscale Air Quality (CMAQ) modelling system (1 km2 resolution). Modelled NO2 concentrations were compared with observations through a series of statistics. Although there are no remarkable differences between both model runs, the results suggest that HBEFA may overestimate traffic emissions. However, the results are strongly influenced by methodological issues and limitations of the traffic model. This study was useful to provide a first alternative estimate to the official emission inventory in Madrid and to identify the main features of the traffic model that should be improved to support the application of an emission system based on “real world” emission factors.

Vista de Almenara, tomada de Murviedro [Material gráfico] = Vue d'Almenara prise de la route de Murviedro=View of Almenara, taken from the road of Murviedra [sic]

Estampada en la misma hoja con dos grabados del Lago de Almenara y con huella de plancha común (53'5 x 31'5 cm)

Vista de Valencia tomada del camino que conduce al Grao [Material gráfico] = Vue de Valence prise sur le chemin qui conduit au Grao = View of Valencia taken on the road leading to the Grao

Estampado junto a "Vista de Valencia tomada de la entrada de la Alameda"

Burma, 1824 (Image 1 of 4) (Raster Image)

This layer is a georeferenced raster image of the historic paper map entitled: Geographical sketch of the Burmese Empire, drawn by H. Hamilton ; compiled at the office of the Surveyor General of India. It was published by Published at the Asiatic Lithographic Press in 1825. Scale [ca. 1:1,010,000]. This layer is image 1 of 4 total images of the four sheet source map, representing the north portion of the map. Covers Burma and bordering portions of India, Bangladesh, Thailand, and China.The image inside the map neatline is georeferenced to the surface of the earth and fit to the Asia North Lambert Conformal Conic coordinate system. All map collar and inset information is also available as part of the raster image, including any inset maps, profiles, statistical tables, directories, text, illustrations, index maps, legends, or other information associated with the principal map. This map shows features such as drainage, roads, cities and other human settlements, fortifications, territorial boundaries, shoreline features, ground cover, temples, and more. Relief shown by hachures. Includes also glossary and notes on the construction of the sketch and table of estimated road distances.This layer is part of a selection of digitally scanned and georeferenced historic maps from the Harvard Map Collection. These maps typically portray both natural and manmade features. The selection represents a range of originators, ground condition dates, scales, and map purposes.

Burma, 1824 (Image 2 of 4) (Raster Image)

This layer is a georeferenced raster image of the historic paper map entitled: Geographical sketch of the Burmese Empire, drawn by H. Hamilton ; compiled at the office of the Surveyor General of India. It was published by Published at the Asiatic Lithographic Press in 1825. Scale [ca. 1:1,010,000]. This layer is image 2 of 4 total images of the four sheet source map, representing the central north portion of the map. Covers Burma and bordering portions of India, Bangladesh, Thailand, and China.The image inside the map neatline is georeferenced to the surface of the earth and fit to the Asia North Lambert Conformal Conic coordinate system. All map collar and inset information is also available as part of the raster image, including any inset maps, profiles, statistical tables, directories, text, illustrations, index maps, legends, or other information associated with the principal map. This map shows features such as drainage, roads, cities and other human settlements, fortifications, territorial boundaries, shoreline features, ground cover, temples, and more. Relief shown by hachures. Includes also glossary and notes on the construction of the sketch and table of estimated road distances.This layer is part of a selection of digitally scanned and georeferenced historic maps from the Harvard Map Collection. These maps typically portray both natural and manmade features. The selection represents a range of originators, ground condition dates, scales, and map purposes.

Burma, 1824 (Image 3 of 4) (Raster Image)

This layer is a georeferenced raster image of the historic paper map entitled: Geographical sketch of the Burmese Empire, drawn by H. Hamilton ; compiled at the office of the Surveyor General of India. It was published by Published at the Asiatic Lithographic Press in 1825. Scale [ca. 1:1,010,000]. This layer is image 3 of 4 total images of the four sheet source map, representing the central south portion of the map. Covers Burma and bordering portions of India, Bangladesh, Thailand, and China.The image inside the map neatline is georeferenced to the surface of the earth and fit to the Asia North Lambert Conformal Conic coordinate system. All map collar and inset information is also available as part of the raster image, including any inset maps, profiles, statistical tables, directories, text, illustrations, index maps, legends, or other information associated with the principal map. This map shows features such as drainage, roads, cities and other human settlements, fortifications, territorial boundaries, shoreline features, ground cover, temples, and more. Relief shown by hachures. Includes also glossary and notes on the construction of the sketch and table of estimated road distances.This layer is part of a selection of digitally scanned and georeferenced historic maps from the Harvard Map Collection. These maps typically portray both natural and manmade features. The selection represents a range of originators, ground condition dates, scales, and map purposes.

Burma, 1824 (Image 4 of 4) (Raster Image)

This layer is a georeferenced raster image of the historic paper map entitled: Geographical sketch of the Burmese Empire, drawn by H. Hamilton ; compiled at the office of the Surveyor General of India. It was published by Published at the Asiatic Lithographic Press in 1825. Scale [ca. 1:1,010,000]. This layer is image 4 of 4 total images of the four sheet source map, representing the south portion of the map. Covers Burma and bordering portions of India, Bangladesh, Thailand, and China.The image inside the map neatline is georeferenced to the surface of the earth and fit to the Asia North Lambert Conformal Conic coordinate system. All map collar and inset information is also available as part of the raster image, including any inset maps, profiles, statistical tables, directories, text, illustrations, index maps, legends, or other information associated with the principal map. This map shows features such as drainage, roads, cities and other human settlements, fortifications, territorial boundaries, shoreline features, ground cover, temples, and more. Relief shown by hachures. Includes also glossary and notes on the construction of the sketch and table of estimated road distances.This layer is part of a selection of digitally scanned and georeferenced historic maps from the Harvard Map Collection. These maps typically portray both natural and manmade features. The selection represents a range of originators, ground condition dates, scales, and map purposes.

Majorca Island, Minorca Island, and Isla de Ibiza, Spain, ca. 1756 (Raster Image)

This layer is a georeferenced raster image of the historic paper map entitled: Carte des l'isles de Maiorque, Minorque et d'Yvice, gravée par Matthieu Seutter, Geographe de S.M. Imper. et Cathol. ; Tob. Conr. Lotter, sc. It was published by M. Seutter ca. 1756. Scale [ca. 1:560,000]. Covers Majorca Island, Minorca Island, and Isla de Ibiza, Spain. Map in French.The image inside the map neatline is georeferenced to the surface of the earth and fit to the European Datum 1950, Universal Transverse Mercator (UTM) Zone 31N projected coordinate system. All map collar and inset information is also available as part of the raster image, including any inset maps, profiles, statistical tables, directories, text, illustrations, index maps, legends, or other information associated with the principal map. This map shows features such as drainage, cities and other human settlements, road, shoreline features, and more. Relief shown by hachures and pictorially. Includes 2 insets, one showing the islands and the western part of the Mediterranean, and, "Plan du port et ville de Mahon, du Fort. St. Philippe et ses fortifications."This layer is part of a selection of digitally scanned and georeferenced historic maps from the Harvard Map Collection. These maps typically portray both natural and manmade features. The selection represents a range of originators, ground condition dates, scales, and map purposes.

British Isles, 1818 (Raster Image)

This layer is a georeferenced raster image of the historic paper map entitled: Cary's reduction of his six sheet map of the British Isles : comprehending the whole of the turnpike roads : with the great rivers and the course of the different navigable canals, also the market and borough towns and villages adjoining the road : to which is added the actual distance from one market town to another ; with the exact admeasurement to each from the metropolis : drawn from the most recent surveys. It was published by J. Cary in 1818. Scale [ca. 1:424,000].The image inside the map neatline is georeferenced to the surface of the earth and fit to the Europe Lambert Conformal Conic coordinate system. All map collar and inset information is also available as part of the raster image, including any inset maps, profiles, statistical tables, directories, text, illustrations, index maps, legends, or other information associated with the principal map. This map shows features such as drainage, cities and other human settlements, territorial boundaries, shoreline features, roads, and more. Relief shown by hachures. Includes inset: The Shetland Isles.This layer is part of a selection of digitally scanned and georeferenced historic maps from the Harvard Map Collection. These maps typically portray both natural and manmade features. The selection represents a range of originators, ground condition dates, scales, and map purposes.

World Map Showing Routes of Exploration and Discovery, 1850 (Raster Image)

This layer is a georeferenced raster image of the historic paper map entitled: The world, on Mercator's projection, by David H. Burr. It was published by J. Haven, 86 State St. in 1850. Scale [ca. 1:74,000,000]. The image inside the map neatline is georeferenced to the surface of the earth and fit to a non-standard 'Mercator' projection with the central meridian at 90 degrees west. All map collar and inset information is also available as part of the raster image, including any inset maps, profiles, statistical tables, directories, text, illustrations, index maps, legends, or other information associated with the principal map. "Showing the different routes to California, and distance by each, routes of different navigators, route of the contemplated Pacific R. Road, distances to China, Europe, & c." This map shows features such as drainage, cities and other human settlements, shoreline features, and more. Relief shown by abbreviated form lines. Includes historical notes, ship ill., tables of distances, "Nicaragua route" distances, and text on "Pacific Railroad." Insets: Map of the Nicaragua route -- Map of the proposed rail road from St. Louis to the bay of St. Francisco. This layer is part of a selection of digitally scanned and georeferenced historic maps from the Harvard Map Collection and the Harvard University Library as part of the Open Collections Program at Harvard University project: Organizing Our World: Sponsored Exploration and Scientific Discovery in the Modern Age. Maps selected for the project correspond to various expeditions and represent a range of regions, originators, ground condition dates, scales, and purposes.

New England Railroads, 1838 (Raster Image)

This layer is a georeferenced raster image of the historic paper map entitled: Sketch of the states of Massachusetts, Connecticut, and Rhode Island, and parts of New Hampshire & New York exhibiting the several rail road routes completed, constructing, chartered & contemplated : published by order of the Legislature of Massachusetts, drawn by A. Kennedy ; engraved by Morse & Tuttle. It was published in 1838. Scale [ca. 1:700,000]. The image inside the map neatline is georeferenced to the surface of the earth and fit to the USA Contiguous Albers Equal Area Conic projection (Meters). All map collar and inset information is also available as part of the raster image, including any inset maps, profiles, statistical tables, directories, text, illustrations, or other information associated with the principal map. This map shows features such as roads, completed, chartered, and contemplated railroads, drainage, state and county boundaries, and more. This layer is part of a selection of digitally scanned and georeferenced historic maps of New England from the Harvard Map Collection. These maps typically portray both natural and manmade features. The selection represents a range of regions, originators, ground condition dates, scales, and purposes.

Troy & Greenfield Railroad, New York and New England, ca. 1855 (Raster Image)

This layer is a georeferenced raster image of the historic paper map entitled: Map of the Troy & Greenfield Rail Road and its connections, [by] A.F. Edwards, chief engineer. It was published ca. 1855 by B.W. Thayer & Co.'s Lith. Scale not given. Covers Vermont, New Hampshire, Massachusetts, Connecticut, Rhode Island, and portions of Maine and New York.The image inside the map neatline is georeferenced to the surface of the earth and fit to the USA Contiguous Albers Equal Area Conic projection (Meters). All map collar and inset information is also available as part of the raster image, including any inset maps, profiles, statistical tables, directories, text, illustrations, or other information associated with the principal map. This map shows features such as railroads completed, chartered and under construction, drainage, selected cities, towns, and villages, state and county boundaries, and more. Relief shown by hachures.This layer is part of a selection of digitally scanned and georeferenced historic maps of New England from the Harvard Map Collection. These maps typically portray both natural and manmade features. The selection represents a range of regions, originators, ground condition dates, scales, and purposes.