In-sensor Computing ANN Capacitive Sensors

TL;DR

This paper introduces a novel in-sensor computing circuit based on capacitance for neural network applications, enabling efficient, low-power ANN classifiers and autoencoders directly within sensors.

Contribution

It presents a new capacitance-based MAC circuit that integrates neural network functionalities into sensor hardware, enhancing efficiency and reducing power consumption.

Findings

High efficiency in capacitive ANN image sensors

Lower power consumption compared to traditional methods

Effective implementation of classifiers and autoencoders

Abstract

This letter proposes an in-sensor computing multiply-and-accumulate (MAC) circuit based on capacitance. The MAC circuits can constitute an artificial neural network(ANN) layer and be operated as ANN classifiers and autoencoders. The proposed circuit is a promising scheme for capacitive ANN image sensors, showing competitively high efficiency and lower power.