PROJECT TITLE :

Visual saliency on networks of neurosynaptic cores

ABSTRACT:

Identifying fascinating or salient regions in a picture plays an important role for multimedia search, object tracking, active vision, segmentation, and classification. Existing saliency extraction algorithms are implemented using the conventional von Neumann computational model. We have a tendency to propose a bottom-up model of visual saliency, impressed by the primate visual cortex, which is compatible with TrueNorth-an occasional-power, brain-inspired neuromorphic substrate that runs massive-scale spiking neural networks in real-time. Our model uses color, motion, luminance, and form to identify salient regions in video sequences. For a three-color-channel video with 240 136 pixels per frame and 30 frames per second, we tend to demonstrate a model utilizing three million neurons, that achieves competitive detection performance on a publicly on the market dataset whereas consuming 200 mW.