header image
Home arrow Project Activities
Project Activities Print E-mail
1. The data reconstruction over the actual flows of the goodses will be carried out through specific surveyings among personnel providing services of transport and logistics. The main evolutionary dynamics of the exchanges, for the merceologic category, and for the flow direction, will be characterized by means of the statistical analysis of the historical series; this will concur to characterize trend and ciclicity of the flows.

2. Forecasts of the traffic flows will be carried out to short, medium and long period, on the base of predictive models represented by applications of parametric and non parametric models of regression.

3. The study and development of intelligent and advanced techniques in the intermodal transport for the management, location and optimization of the available resources. In particular, we intend to study and develop data mining algorithms in order to supply a support to the efficient management of the resources at disposal. Moreover, the problem of the optimization of the transports will be faced, with particular regard to pattern recognition and machine learning models.

4. The definition and planning of modules for video-surveillance systems of quarter generation. We intend to study automatic mechanisms of localization and tracking in controlled environments and mechanisms of integration of the data extracted from images/videos. To such aim, we intend to develop innovative  methodologies of image analysis based on the shape, color and texture analysis in order to recognize vehicles at access crosses, to implement visual attention mechanisms on meaningful areas of the captured scene, to detect object motion, to analyze the posture, to find anomalous and/or dangerous situations.

5. The study and the realization of recognition and tracking systems of motion based on attentive vision and multi-modal biometrical systems in real application contexts. The aim is the realization of a system for the control of the safe accesses in uncontrolled environments. Being based on actual research findings, and in order to obtain accuracy and advanced reliability levels, we intend to adopt attentive vision models for the analysis of dynamic scenes; this with the aim of the object recognition and tracking and the fusion and integration techniques of similarity measures obtained from single sensory modalities (mulit-modal fusion) or from various algorithms applied over the same sensorial data (multi-algorithmic fusion) in biometrics.

Login Form

Lost Password?
No account yet? Register
Kick off 2 february 2008
Who's Online