Human behaviour analysis

The group is starting a new research focused in human behavior analysis in video sequences with multiple targets. A robust tracking system is needed to perform the analysis. Two issues have to be solved: target collisions and occlusions. A collision occurs when two or more targets interact within the scene. The system have to control this situations, in order to avoid possible exchange of identifiers. We have developed a full multiple-target tracking system that can deal with people collisions and occlusions. The system is also able to be trained to detect other objects of interest.
We are currently working on the target path detection and characterization. The system includes a module for abnormal behavior detection that can work with or without any a priori information about the environment. A module of detection of usual paths, including frequency information, is also developed.

Public databases

BARD Dataset

The BARD dataset is a set of videos used for human behavioral analysis and recognition. The dataset currently contains four video sequences. The captures are taken place outdoors, under an uncontrolled scenario.

If you are interested in using the BARD dataset, please send an email to: mortega@udc.es and you will receive an authentication password to access the dataset. This is intended for statistical purposes only, no private data or fee is required. If you use this image set for your work, please include a reference of BARD in it. If you use this dataset for your works, please include a reference of BARD in it by citing the original papers where the database was introduced:

  • B. Cancela, M. Ortega, M. G. Penedo, J. Novo, N. Barreira, «On the Use of a Minimal Path Approach for Target Trajectory Analysis», Pattern Recognition, 46 (7), 2015 – 2027, 2013.
  • B. Cancela, M. Ortega, M. G. Penedo, «Multiple Human Tracking System for Unpredictable Trajectories», Machine Vision and Applications, 25(2), 511-527, 2014.

Main publications

  • B. Cancela, A. Iglesias, M. Ortega, M. G. Penedo, «Unsupervised Trajectory Modelling using Temporal Information via Minimal Paths», IEEE Conference on Computer Vision and Pattern Recognition (CVPR 2014), Columbus, Ohio, June 2014.
Abstract
In this work a methodology for the classification of retinal feature points is applied to a biometric system. This system is based in the extraction of feature points, namely bifurcations and crossovers as biometric pattern. In order to compare a pattern to other from a known individual a matching process takes place between both points sets. That matching task is performed by finding the best geometric transform between sets, i.e. the transform leading to the highest number of matched points. The goal is to reduce the number of explored transforms by introducing the previous characterisation of feature points. This is achieved with a constraint avoiding two differently classified points to match. The empirical reduction of transforms is about 20%.
  • B. Cancela, M. Ortega, M. G. Penedo, «Multiple Human Tracking System for Unpredictable Trajectories», Machine Vision and Applications, 25(2), 511-527, 2014.
Abstract
Biometrics refer to identity verification of individuals based on some physiologic or behavioural characteristics. The typical authentication process of a person consists in extracting a biometric pattern of him/her and matching it with the stored pattern for the authorized user obtaining a similarity value between patterns. In this work an efficient method for persons authentication is showed. The biometric pattern of the system is a set of feature points representing landmarks in the retinal vessel tree. The pattern extraction and matching is described. Also, a deep analysis of similarity metrics performance is presented for the biometric system. A database with samples of retina images from users on different moments of time is used, thus simulating a hard and real environment of verification. Even in this scenario, the system allows to establish a wide confidence band for the metric threshold where no errors are obtained for training and test sets.
  • B. Cancela, M. Ortega, M. G. Penedo, J. Novo, N. Barreira, «On the Use of a Minimal Path Approach for Target Trajectory Analysis», Pattern Recognition, 46 (7), 2015 – 2027, 2013.
Abstract
Biometric features have been studied in order to be applied to authentication and identification systems due to its reliability. Among others, the retinal vessel tree have been proposed as a vessel pattern for personal authentication applications, since it is almost impossible to forge. In this kind of systems, the retinal vessel tree is computed from the retinal image, and then a registration process is made. Although reliable and remarkable results have been obtained in this vessel pattern-based system, the required computation effort is quite high, particularly to compute and extract the vessel tree. In this paper, a pixel parallel approach is proposed to tackle with the retinal vessel tree extraction in order to be used in a personal retinal authentication system, regarding the computation speed. An analysis has been made to test the performance of our obtained tree with the original proposal.
  • B. Cancela, M. Ortega, A. Fernández, M. G. Penedo, «Hierarchical framework for robust and fast multiple-target tracking in surveillance scenarios», Expert Systems with Applications, 40, 1116 – 1131, 2013.
Abstract
Traditional authentication (identity verification) systems, employed to gain access to a private area in a building or to data stored in a computer, are based on something the user «has» (an authentication card, a magnetic key) or something the user «knows» (a password, an identification code). But emerging technologies allow for more reliable and comfortable for the user, authentication methods, most of them based in biometric parameters. Much work could be found in literature about biometric based authentication, using parameters like iris, voice, fingerprint, face characteristics, and others. In this work a novel authentication method is presented, and first results obtained are shown. The biometric parameter employed for the authentication is the retinal vessel tree, acquired through a retinal angiography. It has already been asserted by expert clinicians that the configuration of the retinal vessels is unique for each individual and that it does not vary in his life, so it is a very well suited identification characteristic. Before the verification process can be executed, a registration step is needed to align both the reference image and the picture to be verified. A fast and reliable registration method is used to perform that step, so that the whole authentication process takes very little time.