A Multi-target Detection and Tracking Algorithm

Clc Number:


Fund Project:

Research on Autonomous Path Planning and Obstacle Avoidance of Mobile Robot Based on Binocular Vision

  • Article
  • |
  • Figures
  • |
  • Metrics
  • |
  • Reference
  • |
  • Related
  • |
  • Cited by
  • |
  • Materials
  • |
  • Comments

    This paper proposes a multi-target detection and tracking algorithm based on the YOLOv3 architecture to achieve target detection and core-related filtering algorithms to achieve target tracking. First, use the trained YOLOv3 network to obtain the location of the target in the video, and assign ID to each target; second, input multiple targets in parallel to the tracking module based on nuclear correlation filtering for target tracking; then, determine whether the start correction is satisfied If the conditions of the strategy are met, the results of the detection module are used to modify the results of the tracking module; finally, the tracking results are used to update the kernel correlation filter model. Through experimental analysis, the overall average success rate of the algorithm proposed in this paper has reached 81.2%, which verifies the effectiveness of the algorithm.

    Cited by
Get Citation
Article Metrics
  • Abstract:
  • PDF:
  • HTML:
  • Cited by:
  • Received:April 21,2021
  • Revised:April 22,2021
  • Adopted:March 31,2022
  • Online:
  • Published: