In the fascinating world of ants, ant colonies are described as eusocial, i.e. reaching the highest levels of organization in animal sociality.
Their collaborative capabilities is one of the major keys of such a success. Through collaboration, ants achieve wonders hardly conceivable
when considering their individual size and strength. By analogy, we propose a novel model of camera network, compatible with cameras of very
low specificatio
ns (e.g., low resolution, limited processing performance):Ant-Cam, building then our platform LobNet.
In camera networks, we classically try to optimize the hardware parameters, including position,
orientation, field of view (FOV), zoom, focus and resolution at the camera level. But this is also
true for software components (e.g., detection algorithm) as well as topological components
(e.g., number of necessary cameras to cover the task space). With our Ant-Cam, we choose to
focus on the Ant-Cams’ interactions to increase the
performance of the system despite their limited capabilities and non-optimal parameters.
For our silly cameras, we choose to use a mouse sensor giving very small images (30×30!!!) in grayscale
with 5fps processed by a max 10 FPGA from Altera. To highlight theirs talkativeness capacity, we choose
SmartMesh IP protocol from Linear technology.
Mouse sensor and an image….
Lobna BenKhelifa,
Padawan DREAM