A new session of the series of seminars for the AMTC staff, in which the projects of our researchers are presented and discussed, was held on Thursday, June 28th. This was the turn of Dr. Leonardo Cament, currently doing a postdoctorate in our Center.
Dr. Cament presented his work: “A Multi-Sensor, Gibbs Sampled, Implementation of the Multi-Bernoulli Poisson Filter”, which has applications in computer vision and multi-target tracking (to simultaneously detect several objects, recognize them and track them).
In his presentation, Dr. Cament stated that tests have been done using sensors like radar and lasers, giving satisfactory results in tracking objects that simultaneously enter into the visual field of a machine. However, there’s still progress to be done in cases such as elements that dissapear from the visual field without exiting it, undetected elements or taking one moving element for two (Dr. Cament mentioned a case in which the computer saw a walking person who was moving his arms, and thought there were two separated elements: the person and the arm).
Dr. Cament highlighted a practical use for this system: automatic detection of people entering a restricted or hazardous area, specially in mining environments.