An Event-Driven Multi-Kernel Convolution Processor Module for Event-Driven Vision Sensors Luis Camuñas-Mesa, Carlos Zamarreño-Ramos

Abstract

Event-Driven vision sensing is a new way of sensing visual reality in a frame-free manner. This is, the vision sensor (camera) is not capturing a sequence of st ill frames, as in conven- tional video and computer vision systems. In Event-Driven sen- sors each pixel autonomously and asynchronously decides when to send its address out. This way, t he sensor output is a continuous stream of address events representing reality dynamically continu- ously and without constraining to frames. In this paper we present an Event-Driven Convolutio n Module for computing 2D convolu- tions on such event streams. The Convolution Module has been designed to assemble many of them for building modular and hi- erarchical Convolutio nal Neural Networks for robust shape and pose invariant object recognit ion. The Convolution Module has multi-kernel capability. This is, it will select the convolution kernel depending on the origin of the event. A proof-of-concept test proto- type has been fabricated in a 0.35 m CMOS process and extensive experimental results are provided. The Convolution Processor has also been comb ined with an Event-Driven Dynamic Vision Sensor (DVS) for high-speed recognition examples. The chip can discrim- inate propellers rotating at 2 k re volutions per second, detect sym- bols on a 52 card deck when browsing all cards in 410 ms, or detect and follow the center of a phosphor oscilloscope trace rotating at 5K H z .