Localization of Planar Objects in Images Bearing Repetitive Patterns.

Device and automatic procedure for establishing the sex of faces appearing in images.

Dependencies between facial attributes given the appearance. Recent works report significant drops in performance for state-of-the-art gender classifiers when evaluated "in the wild", i.e. with uncontrolled demography and environmental conditions. By considering the relation between gender and pose attributes we also avoid the use of computation-ally expensive and fragile face alignment procedures. In the experiments we confirm the existence of dependencies among gender, age and pose facial attributes and prove that we can improve the performance and robustness of gender classifiers by exploiting these dependencies.

Facial age estimation . We developed a "light weight" face age estimation algorithm using PCA+LDA and KNN regression.

PCA+LDA face gender recognition. We developed gender classification algorithm using an simple holistic approach. We got 93% accuracy on FERET database (5-fold cross validation) and an speed comparable to the fastest gender recognition algorithms. This algorithm is based on PCA+LDA with proper cross-validation of PCA dimensions.

Face expressions recognition. We used our 2D appearance tracking algorithm [BMVC'2006] to track the face. We developed a user independent manifold of face expressions using PCA+LDA and then we classified the expressions using KNN.

Appearance-based reanimation from expressions estimation. We developed a way of doing 3D face reanimation based on appearance-based technicques.

Multiclass cost-sensitive boosting: BAdaCost. We introduce a cost-sensitive multi-class Boosting algorithm (BAdaCost) based on a generalization of the Boosting margin, termed multi-class cost-sensitive margin. It can be used, for example, to address the class imbalance by the introduction of a cost matrix that weighs more hevily the costs of confused classes. Other interesting uses are multi-view object detection.

Speeding-up homography estimation in mobile devices. We introduce a procedure for reducing the number of samples required for fitting a homography to a set of noisy correspondences using a random sampling method. This is achieved by means of a geometric constraint that detects invalid minimal sets.

Efficient 3d nonrigid tracking. Efficient incremental image alignment is a topic of renewed interest in the computer vision community because of its applications in model fitting and model-based object tracking. We are working in efficient solutions to the 3D tracking of a head performing face expressions under changing illumination conditions.

Efficient appearance-based tracking with illumination changes and face expressions. separates facial expressions from illumination variations. The appearance of a face is represented by the addition of two independent linear subspaces modelling facial expressions and illumination. This simple model enables us to train the system with no manual intervention. We also introduce an efficient procedure for fitting this model, which can be used for tracking a human face in real-time.

Efficient appearance-based tracking We have developed an efficient way of minimizing the eigentracking for nonrigid motion estimation. It is based on the precomputation of motion templates to save on-line computation. It allows as to estimate appearance (PCA coefficients) and motion in real-time.

SSD based 3D tracking. We have developed an algorithm for tracking a rigid object based on a piecewise planar model. The tracking is performed using a single incremental SSD-based tracker. The main feature of the approach presented is that it can track a rigid set of arbitrarily small patches all of which could not be individually tracked.

SSD based tracking. Planar tracking can be used for face tracking. We have extended a well known framework for planar tracking (see [Hager98]) with a projective motion model. Additionally, using a calibrated camera, it is possible to estimate the 3D pose of the planar object. We have also developed a procedure to select the most informative pixels of the target template image for faster tracking.

Colour Based Tracking. Tracking using colour is difficult when sudden light colour changes take place. We have extented a well known colour constancy algorithm, Grey World, to deal with such situations. The result is more robust than widely used RGB-normalisation, although it is not perfect either.

Robust face tracking. All the algorithms based on a simple visual cues fail in some circumstances. The key idea is to use this simple algorithms together in order to get robustness. All this algorithms should be "orthogonal" in the sense of having different fail conditions.

Building Facade Localization. We developed new algorithms for real-time planar building facades localization. This is not an easy problem given the repetitive structures in the building texture. Work funded by The Graffer.

Our algorithms, implemented by the Computer Vision engineers at The Graffter, are able to run at 6-10 fps in a Google Nexus 5 on Android.

OCR for ID documents on video. We developed and implemented our own algorithms for the OCR of spanish ID documents. The funding came from Work funded by CAB Magazine Online S.L. (an Spanish company).

The goal of the project was to read the 3 OCR lines of the spanish ID documents in short videos. Other countries have only 2 OCR lines. Our software is valid for this other documents also with minimal customization. In this project we developed a full prototype using C++ and OpenCV on GNU/Linux. The prototype is fully functional (10 fps on an Intel Core i5). It makes one mistake on average per ID document (in our limited test cases of few IDs). Unfortunately the company gave up on the project before the full validation of the prototype so we have the software and the project available to any interested company :-)

Facial analysis for video analythics. We developed a set of low computational requirements face analysis algorithms for, Visión Artificial Desarrollos de I+D, an Spanish company. The algorithms were integrated in a product called vAudience for Digital Signage audience measurement.