Author Archives: tama

Primera batería de fotosFirst photograph battery

Ya hemos comenzado a tomar fotos de la ciudad de Zaragoza para entrenar al sistema de reconocimiento automático. Las fotografías se etiquetan automáticamente con su posición GPS en el momento en que se toman. Se van almacenando en una web del proyecto:

Este es el mapa general:
mapa_general

Detalle de la Plaza del Pilar:
mapa_la_seo

La zona de La Seo:
mapa_la_seo_detalla_imagen

Algunas de las fotografías de la galería:
1_galeria

galeria_lonja

We have started the battery of photograps around the city. This phase is necessary in order to train the visual recognition system. The photographs are automatically labeled with their GPS position, and they are stored in a web:

This is the general map:
mapa_general

A detail of Plaza del Pilar:
mapa_la_seo

The zone of Plaza de La Seo:
mapa_la_seo_detalla_imagen

Some of the photos in the gallery:
1_galeria

galeria_lonja

Aceptado un artículo en CEIG 2014Article accepted in CEIG 2014

El artículo “Improving depth estimation using superpixels” ha sido aceptado para su presentación en el Congreso Español de Informática Gráfica, CEIG 2014. Los autores son Ana B. Cambra, Adolfo Muñoz, Ana C. Murillo, José J. Guerrero y Diego Gutiérrez.

Este es el abstract:

This work is focused on assigning a depth label to each pixel in the image. We consider off-the-shelf algorithms that provide depth information from multiple views or depth information directly obtained from RGB-d sensors. Both of them are common scenarios of a well studied problem where many times we get incomplete depth information. Then, user interaction becomes necessary to finish, improve or correct the solution for certain applications where accurate and dense depth information for all pixels in the image is needed. This work presents our approach to improve the depth assigned to each pixel in an automated manner. Our proposed pipeline combines state-of-the art methods for image superpixel segmentation and energy minimization. Superpixel segmentation reduces complexity and provides more robustness to the labeling decisions. We study how to propagate the depth information to incomplete or inconsistent regions of the image using a Markov Random Field (MRF) energy minimization framework. We propose and evaluate an energy function and validate it together with the designed pipeline. We present a quantitative evaluation of our approach with different variations to show the improvements we can obtain. This is done using a publicly available stereo dataset that provides ground truth information. We show additional qualitatively results, with other tests cases and scenarios using different input depth information, where we also obtain significant improvements on the depth estimation compared to the initial one. CEIG2014

The article“Improving depth estimation using superpixels” has been acceped in Congreso Español de Informática Gráfica, CEIG 2014. The authors are Ana B. Cambra, Adolfo Muñoz, Ana C. Murillo, José J. Guerrero y Diego Gutiérrez.

Abstract:

This work is focused on assigning a depth label to each pixel in the image. We consider off-the-shelf algorithms that provide depth information from multiple views or depth information directly obtained from RGB-d sensors. Both of them are common scenarios of a well studied problem where many times we get incomplete depth information. Then, user interaction becomes necessary to finish, improve or correct the solution for certain applications where accurate and dense depth information for all pixels in the image is needed. This work presents our approach to improve the depth assigned to each pixel in an automated manner. Our proposed pipeline combines state-of-the art methods for image superpixel segmentation and energy minimization. Superpixel segmentation reduces complexity and provides more robustness to the labeling decisions. We study how to propagate the depth information to incomplete or inconsistent regions of the image using a Markov Random Field (MRF) energy minimization framework. We propose and evaluate an energy function and validate it together with the designed pipeline. We present a quantitative evaluation of our approach with different variations to show the improvements we can obtain. This is done using a publicly available stereo dataset that provides ground truth information. We show additional qualitatively results, with other tests cases and scenarios using different input depth information, where we also obtain significant improvements on the depth estimation compared to the initial one. CEIG2014

Publicados dos artículosTwo articles have been published

Las versiones definitivas de dos artículos han sido publicadas recientemente en dos revistas internacionales “open source”. Se trata de:

The definitive versions of two papers in open source journals have just been published, namely:

Café navideñoChristmas celebration

Hoy hemos tenido la celebración Navideña en el edificio I+D, con un café en el que han participado los otros grupos que trabajan aquí.

navidad_1Today we have organized a Christmas celebration with the other people that work in the R&D building in Walqa.

navidad_1

Aceptado un artículo sobre tráfico de videojuegosAccepted article about a traffic model for a video game

Hoy se ha aceptado para su publicación en la revista International Journal of Computer Games Technologyel artículo Analyzing the effect of TCP and server population on massively multiplayer games,” escrito en colaboración con dos investigadores de la Universidad de Zagreb. Los autores son Mirko Suznjevic, Jose Saldana, Maja Matijasevic, Julián Fernández-Navajas, y José Ruiz-Mas. El artículo se puede leer aquí.

El artículo propone un modelo de tráfico para el videojuego “World of Warcraft”, según las horas del día y la actividad realizada por el jugador.

day_wowToday the article Analyzing the effect of TCP and server population on massively multiplayer games, has been accepted for publication in the International Journal of Computer Games Technology. It has been authored by researchers from University of Zagreb and University of Zaragoza: Mirko Suznjevic, Jose Saldana, Maja Matijasevic, Julián Fernández-Navajas, y José Ruiz-Mas. The provisional text can be found here.

The article proposes a traffic model for “World of Warcraft”, which varies according to the hour of the day and the activities deployed by the player.

day_wow

Primeros resultados de detección y trackingFirst detecting and tracking results

Se están consiguiendo los primeros resultados de detección y tracking de formas. Como primer ejemplo se está utilizando el Cubo de la Plaza de la Seo de Zaragoza, en el que se está insertando un cuadro de Francisco de Goya.

Animacion_Warp3_small

Aquí se puede ver un vídeo: Vídeo Goya_smallWe are obtaining the first results about the detection and tracking of different objects. As a first example we are using the Cube of Plaza de La Seo in Zaragoza, and we are adding a drawing of Francisco de Goya on one of its faces.

Animacion_Warp3_small

You can see a video here: Vídeo Goya_small

Reunión del Comité CientíficoScientific Committee Meeting

Hoy hemos tenido una reunión del Comité Científico del Proyecto TAMA. La reunión ha tenido lugar en nuestro edificio de Walqa. Hemos mostrado nuestros resultados hasta la fecha, especialmente en la extracción de características de una imagen, el tracking de un objeto en una imagen en movimiento, y también en la mejora de un programa para la asignación de profundidad en superpíxeles.

IMG_2608Today we have had a meeting of the Scientific Committee of TAMA Project in our building in Walqa. We have shown the results so far, especially regarding the extraction of characteristics from an image, the tracking of an object in a moving image, and finally the improvement of an application for assigning depth to super-pixels.

IMG_2608

Publicado un artículo sobre transmisión de vídeoAn article about video transmission has been published

La semana pasada publicamos un artículo en la revista The Scientific World Journal. El título es Video conferences through the Internet: How to Survive in a Hostilehostile Environment“. Estudia y compara dos soluciones comerciales de videoconferencia (Vidyo y Skype), centrándose en los mecanismos que se implementan para funcionar correctamente con condiciones de red difíciles.hostileLast week we got accepted an article in The Scientific World Journal. The title is Video conferences through the Internet: How to Survive in a Hostilehostile Environment“. It studies and compares two commercial video-conference solutions (namely Vidyo and Skype), with a special focus on the mechanisms they implement in order to work properly even in difficult network conditions.hostile