HyperCam: Low-Power Onboard Computer Vision for IoT Cameras

TL;DR

HyperCam introduces an energy-efficient image classification pipeline using hyperdimensional computing, enabling low-power IoT cameras to perform accurate computer vision tasks onboard with minimal resource consumption.

Contribution

It is the first to leverage hyperdimensional computing for low-power onboard image classification in IoT cameras, achieving high accuracy with minimal resource usage.

Findings

Achieves over 93% accuracy on MNIST and Fashion-MNIST datasets.

Maintains low inference latency of 0.08-0.27 seconds.

Uses only 42.91-63 KB flash memory and 22.25 KB RAM.

Abstract

We present HyperCam, an energy-efficient image classification pipeline that enables computer vision tasks onboard low-power IoT camera systems. HyperCam leverages hyperdimensional computing to perform training and inference efficiently on low-power microcontrollers. We implement a low-power wireless camera platform using off-the-shelf hardware and demonstrate that HyperCam can achieve an accuracy of 93.60%, 84.06%, 92.98%, and 72.79% for MNIST, Fashion-MNIST, Face Detection, and Face Identification tasks, respectively, while significantly outperforming other classifiers in resource efficiency. Specifically, it delivers inference latency of 0.08-0.27s while using 42.91-63.00KB flash memory and 22.25KB RAM at peak. Among other machine learning classifiers such as SVM, xgBoost, MicroNets, MobileNetV3, and MCUNetV3, HyperCam is the only classifier that achieves competitive accuracy while maintaining competitive memory footprint and inference latency that meets the resource requirements of low-power camera systems.