Variable Resolution Pixel Quantization for Low Power Machine Vision Application on Edge

TL;DR

This paper introduces a variable resolution pixel quantization method using Hadamard transform in the analog domain to reduce power consumption in edge machine vision, while maintaining high classification accuracy.

Contribution

It proposes a novel analog transformation and quantization algorithm that significantly lowers bits-per-pixel in sensor data without sacrificing CNN classification performance.

Findings

Achieves 90% accuracy on CIFAR-10 with 3-BPP images

Reduces power dissipation in sensor hardware

Demonstrates effectiveness across various image sizes and ADC configurations

Abstract

This work describes an approach towards pixel quantization using variable resolution which is made feasible using image transformation in the analog domain. The main aim is to reduce the average bits-per-pixel (BPP) necessary for representing an image while maintaining the classification accuracy of a Convolutional Neural Network (CNN) that is trained for image classification. The proposed algorithm is based on the Hadamard transform that leads to a low-resolution variable quantization by the analog-to-digital converter (ADC) thus reducing the power dissipation in hardware at the sensor node. Despite the trade-offs inherent in image transformation, the proposed algorithm achieves competitive accuracy levels across various image sizes and ADC configurations, highlighting the importance of considering both accuracy and power consumption in edge computing applications. The schematic of a novel 1.5 bit ADC that incorporates the Hadamard transform is also proposed. A hardware implementation of the analog transformation followed by software-based variable quantization is done for the CIFAR-10 test dataset. The digitized data shows that the network can still identify transformed images with a remarkable 90% accuracy for 3-BPP transformed images following the proposed method.