This page contains pointers to the software and data developed and produced by out team members.

Unsupervised Image Alignment Code

The code package contains several different methods developed by us and others for unsupervised alignment of images and, in particular, aligment of images containing examples from a semantic visual class (motorbikes, cars, faces, etc.)

Bitbucket repository

Unsupervised Gaussian Mixture Models and Bayesian Classification

Matlab functionality for learning finite Gaussian mixtures from sample data and performing data classification with Mahalanobis distance or the Bayesian decision rule. Three variations of the Expectation Maximization algorithm are supported: the basic EM algorithm with covariance fixing, the Figueiredo-Jain clustering algorithm and the greedy EM algorithm which both are unsupervised methods (no need to define the number of Gaussians).

Bitbucket repository

MVPRMATLAB - Machine Vision and Pattern Recognition Matlab Toolbox

A Bitbucket repository full of useful Matlab functions developed during our projects.


DiaRetDB1 is a public database for evaluating and benchmarking diabetic retinopathy detection algorithms. The database contains digital images of eye fundus and expert annotated ground truth for several well-known diabetic fundus lesions (hard exudates, soft exudates, microaneurysms and hemorrhages). The original images and the raw ground truth are both available.

RTMOSAIC - Real-time mosaicing and 3-D reconstruction

RTMOSAIC project developed an online 3D-reconstruction pipeline, fed from a single hand-held camera. Camera ego-motion is tracked using an extended Kalman filter visual simultaneous localization and mapping (EKF-VSLAM). Selected frames from the video are processed by GPU-based multiview stereo (MVS) to produce point clouds with color and normals. Source code and data sets are available for download on the project page.

SimpleGabor - Multiresolution Gabor Feature Toolbox

SimpleGabor is a Matlab toolbox for multiresolution Gabor filtering of 2-D signals (images). The implementation is intended to be as efficient as possible within limitations of Matlab.