Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

CloudEye: A New Paradigm of Video Analysis System for Mobile Visual Scenarios

Huan Cui, Qing Li, Hanling Wang, Yong jiang

TL;DR

CloudEye is a real-time, efficient mobile visual perception system that leverages content information mining on edge servers in a mobile vision system environment equipped with edge servers and coordinated with cloud servers.

Abstract

Mobile deep vision systems play a vital role in numerous scenarios. However, deep learning applications in mobile vision scenarios face problems such as tight computing resources. With the development of edge computing, the architecture of edge clouds has mitigated some of the issues related to limited computing resources. However, it has introduced increased latency. To address these challenges, we designed CloudEye which consists of Fast Inference Module, Feature Mining Module and Quality Encode Module. CloudEye is a real-time, efficient mobile visual perception system that leverages content information mining on edge servers in a mobile vision system environment equipped with edge servers and coordinated with cloud servers. Proven by sufficient experiments, we develop a prototype system that reduces network bandwidth usage by 69.50%, increases inference speed by 24.55%, and improves detection accuracy by 67.30%

CloudEye: A New Paradigm of Video Analysis System for Mobile Visual Scenarios

TL;DR

CloudEye is a real-time, efficient mobile visual perception system that leverages content information mining on edge servers in a mobile vision system environment equipped with edge servers and coordinated with cloud servers.

Abstract

Mobile deep vision systems play a vital role in numerous scenarios. However, deep learning applications in mobile vision scenarios face problems such as tight computing resources. With the development of edge computing, the architecture of edge clouds has mitigated some of the issues related to limited computing resources. However, it has introduced increased latency. To address these challenges, we designed CloudEye which consists of Fast Inference Module, Feature Mining Module and Quality Encode Module. CloudEye is a real-time, efficient mobile visual perception system that leverages content information mining on edge servers in a mobile vision system environment equipped with edge servers and coordinated with cloud servers. Proven by sufficient experiments, we develop a prototype system that reduces network bandwidth usage by 69.50%, increases inference speed by 24.55%, and improves detection accuracy by 67.30%

Paper Structure

This paper contains 22 sections, 3 equations, 15 figures, 3 algorithms.

Table of Contents

  1. Introduction
  2. MOTIVATION and CHALLENGES
  3. Mobile Visual Scenarios
  4. Limitations of Existing Solutions
  5. Design Challenges
  6. OVERVIEW AND DESIGN GUIDELINES
  7. Fast Inference Module
  8. Module Implementation
  9. Appendix Case
  10. Feature Mining Module
  11. Module Implementation
  12. Get Cloud Server Results
  13. Feature Extraction at the Edge Server
  14. Quality Encode Module
  15. Clustering Crops
  16. ...and 7 more sections

Figures (15)

  • Figure 1: System Architecture
  • Figure 2: Proposals Comparison: the subgraph (a) shows all the proposals generated by the original model, and the red boxes in the subgraph (b) are the prior proposals provided by CloudEye's Fast inference module.
  • Figure 3: Feature Mining Module Effect: in subgraph (a), the colored boxes are the targets of the reference frame. In subgraph (b), the colored boxes except the red boxes are objects detected by the edge model, while the red boxes are additional objects detected by CloudEye's Feature Mining Module.
  • Figure 4: Clustering crops: it shows ROI areas of the same frame under different cluster numbers. As the number of clusters increases, there are more discrete ROI areas with more accurate division.
  • Figure 5: Platform: an UAV equipped with NVIDIA JETSON edge server and a HD camera.
  • ...and 10 more figures