Feature Compression for Rate Constrained Object Detection on the Edge

TL;DR

This paper introduces a learnable feature compression method for offloading part of YOLO object detection to edge servers, optimizing detection accuracy and computation time under rate constraints for mobile devices.

Contribution

It proposes a novel, lightweight feature compression approach trained jointly with YOLO to improve detection accuracy at low data rates in edge offloading scenarios.

Findings

Higher detection accuracy at low to medium rates compared to baseline methods.

Significantly reduced computation time on mobile devices with CPU only.

Effective trade-off between data rate, detection performance, and computational efficiency.

Abstract

Recent advances in computer vision has led to a growth of interest in deploying visual analytics model on mobile devices. However, most mobile devices have limited computing power, which prohibits them from running large scale visual analytics neural networks. An emerging approach to solve this problem is to offload the computation of these neural networks to computing resources at an edge server. Efficient computation offloading requires optimizing the trade-off between multiple objectives including compressed data rate, analytics performance, and computation speed. In this work, we consider a "split computation" system to offload a part of the computation of the YOLO object detection model. We propose a learnable feature compression approach to compress the intermediate YOLO features with light-weight computation. We train the feature compression and decompression module together with the YOLO model to optimize the object detection accuracy under a rate constraint. Compared to baseline methods that apply either standard image compression or learned image compression at the mobile and perform image decompression and YOLO at the edge, the proposed system achieves higher detection accuracy at the low to medium rate range. Furthermore, the proposed system requires substantially lower computation time on the mobile device with CPU only.