Javier Yebes, PhD. RobeSafe, UAH

Research Projects

This is framed in the eCall paneuropean system for automatic emergency calls from vehicles in case of accident. The occupant monitoring is in charge of detecting the number of passengers and wirelessly transmitting some images from a small and fish-eye onboard camera under the rear-view mirror. Then, it can also facilitates HOV lane monitoring and special charging depending on the vehicle occupants.

Real-time automated system for video surveillance in shopping malls. It is intended to support security agents duties for detecting suspicious human behaviours as: loitering, unattended cash desk, quick entries or exits. (Journal paper in reviewing process).

Camera monitoring of outdoor parkings is not only for security reasons. Multiple smart and automated services can be provided to society. This is thanks to the current state-of-the-art in computer vision and machine learning. Then, this project will apply innovative techniques for background subtraction, object detection and tracking, applying last advancements in research to a real parking monitoring system. (ongoing project).

Multiple view geometry allows to obtain 3D information from a set of overlapping images. Therefore, amazing 3D scenes can be reconstructed up to a scale from several high resolution images of a place, a building, an object, .... Among others, 3D reconstruction can be employed in the area of photogrammetry and topography, 3D modelling, building restoration, etc. This project brings the latest research advancements in 3D reconstruction to real company products and services. (ongoing project).

Wide angle cameras are placed on a vehicle chassis in order to obtain a surrounding view of the vehicle perimeter. Intended for large military vehicles to enhance perimeter security. A prototype in a small robotic platform was build to validate the approach based on visual features, image registering, undistortion removal and final stitching and blending to panorama creation.

Research collaborations

Driver Monitoring. Advanced Driver Assistance System (ADAS) to detect driver drowsiness based on an onboard camera. Driver eyes are tracked for blinking detection and PERCLOS estimation. Furthermore, this system is benefited from the fusion of vehicle indicators (lateral position, heading error and steering wheel angle). Experiments were carried out on a naturalistic truck simulator and on real roads. Paper.

Vision-based Localization. Visual SLAM system for assisting visually impaired people based on visual features, SFM techniques and 3D scene flow. A stereo camera placed on user's chest is employed to map indoor and outdoor environments and locate the person on them. Several video sequences were recorded by 4 users with different visual impairments. Paper

3D Reconstruction of Legacy Buildings to assist in photogrammetry for their study and renovation, as a less costly alternative to laser approaches. Based on capturing high resolution and overlapping images of the buildings. The 3D point clouds are computed from visual features matching and pairwise image registering. Finally, meshes and textures are computed and projected for final representation.

DARPA Virtual Robotics Challenge. I participated in the VRC competition of the DARPA Robotics Challenge as a member of the SARBOT team. At UAH, we were in charge of the robot perception for visual navigation and obstacle avoidance based on the stereo and lidar sensors of the provided humanoid robot.

Text Detection and Recognition in Traffic Panels. Intelligent vehicles can assist drivers in recognizing the characters in highway traffic panels under challenging conditions. An approach based on panels classification and segmentation, visual description of characters and local dictionaries is successfully tested on real roads. Paper

Research Internships and Fellowships

I spent five months in 2013 as visitor researcher at Visual Inference Group, in the Technische Universität Darmstadt (Germany), under the supervision of Prof. Stefan Roth. I worked on object detection and orientation estimation from images of the KITTI Vision benchmark. It was supported with a 4-months grant from the Spanish Ministry of Education FPU program.

I spent five months between 2011 and 2012 as visitor researcher at Toyota Technological Institute at Chicago (USA), under the supervision of Prof. Raquel Urtasun. I worked on 3D scene understanding, learning and inference with Conditional Random Fields. It was supported with a 3-months grant from the University of Alcalá FPI program and 2-months funding from TTI-C.

In 2011, I was granted for 2.5 years by the Spanish Ministry of Education with 1 of the 950 National fellowships as PhD candidate in the Department of Electronics, UAH.

In 2010, I was granted for 4 years by the University of Alcalá with 1 of the 13 fellowships as PhD candidate in the Department of Electronics, UAH. Effective period: May 2010 - November 2011.

Participation in public and private funded projects

  Smart Driving Applications. Funded by the Spanish Ministry of Education and Competitiveness for the period 2013-2015.

  Robocity2030 II-CM. Funded by the Community of Madrid for the period 2010-2013 and participating in conjunction with the remaining 5 public Universities of Madrid.

  DRIVER-ALERT. Funded by the Spanish Ministry of Competitiveness and Work for the period 2009-2011.

  Robocity2030-CM. Funded by the Community of Madrid for the period 2006-2009 and participating in conjunction with the remaining 5 public Universities of Madrid.
  SmartAparc. Funded by CI3-UAH and started in October 2014.

  Expert Video Surveillance System. Funded by ANOVA IT Consulting for 5 months in 2012.

  Advanced Driver Assistance Systems. Funded by FICOMIRRORS S.A. for two years (2009-2010).