EdgeMA: Model Adaptation System for Real-Time Video Analytics on Edge Devices

TL;DR

EdgeMA is a system that enables real-time video analytics on resource-limited edge devices by detecting data drift and adapting models accordingly, significantly improving accuracy in changing environments.

Contribution

The paper introduces EdgeMA, a novel system combining texture-based domain shift detection and importance weighting for model adaptation on edge devices.

Findings

Significantly improves inference accuracy in real-world video streams.

Effectively detects domain shifts using texture features and Random Forest classifier.

Enhances model robustness against data drift in resource-constrained environments.

Abstract

Real-time video analytics on edge devices for changing scenes remains a difficult task. As edge devices are usually resource-constrained, edge deep neural networks (DNNs) have fewer weights and shallower architectures than general DNNs. As a result, they only perform well in limited scenarios and are sensitive to data drift. In this paper, we introduce EdgeMA, a practical and efficient video analytics system designed to adapt models to shifts in real-world video streams over time, addressing the data drift problem. EdgeMA extracts the gray level co-occurrence matrix based statistical texture feature and uses the Random Forest classifier to detect the domain shift. Moreover, we have incorporated a method of model adaptation based on importance weighting, specifically designed to update models to cope with the label distribution shift. Through rigorous evaluation of EdgeMA on a real-world dataset, our results illustrate that EdgeMA significantly improves inference accuracy.