Latent-Space Inpainting for Packet Loss Concealment in Collaborative Object Detection

TL;DR

This paper introduces a novel latent-space inpainting method based on partial differential equations to recover missing deep features caused by packet loss in collaborative object detection, significantly improving data recovery performance.

Contribution

It presents a new approach leveraging PDE-based inpainting in latent space, advancing the state of the art for handling packet loss in edge-cloud object detection systems.

Findings

PDE-based inpainting effectively recovers missing features.

The method outperforms existing data recovery techniques.

Achieves state-of-the-art results in collaborative object detection.

Abstract

Edge devices, such as cameras and mobile units, are increasingly capable of performing sophisticated computation in addition to their traditional roles in sensing and communicating signals. The focus of this paper is on collaborative object detection, where deep features computed on the edge device from input images are transmitted to the cloud for further processing. We consider the impact of packet loss on the transmitted features and examine several ways for recovering the missing data. In particular, through theory and experiments, we show that methods for image inpainting based on partial differential equations work well for the recovery of missing features in the latent space. The obtained results represent the new state of the art for missing data recovery in collaborative object detection.