Personal tools
You are here: Home Publications Analysis of the local statistics at the centre of fixation during visual scene exploration
About NIFTi
NIFTi is about human-robot cooperation. About teams of robots and humans doing tasks together, interacting together to try and reach a shared goal. NIFTi looks at how the robot could bear the human in mind. Literally. When determining what to do or say next in human-robot interaction; when, and how. NIFTi puts the human factor into cognitive robots, and human-robot team interaction in particular.   Each year, NIFTi evaluates its systems together with several USAR organizations. Rescue personnel teams up with NIFTi robots to carry out realistic missions, in real-life training areas. 

This site uses Google Analytics to record statistics on site visits - see Legal information.


A. Carbone and F. Pirri (2010)

Analysis of the local statistics at the centre of fixation during visual scene exploration

In: Proc. IARP International Workshop on Robotics for risky interventions and Environmental Surveillance, RISE 2010.

We describe how to model the statistics of both the visual input selection and gaze orienting be- haviour of a human observer undertaking a visual ex- ploration task in a given visual scenario. Evidences from neuroscience research prove that complex visual systems have shaped their receptive fields and neural organisation in a continuous adaptation to the visual environment stimuli in which they evolved. This ecology is the basis of our investigation of the human visual behaviour from a real set of gaze tracked points of fixation acquired from a custom designed wearable device. By comparing these set of fovea-centred patches with a randomly chosen set of image patches, extracted from the entire observed scene, we aim at characterising the statistical properties and regularities of the selected visual input. Samples from a human observers are collected both in free-viewing and surveillance-like tasks. In this work we suggest that a generative model of the visual input emerging from the recorded scan-path can be used to model a set of feature detectors for the design of an artificial visual system.