Development of a real-time object-tracking system using Raspberry Pi

Abstract

Object tracking uses computerized algorithms to locate and track targets automatically without human intervention. Applying object-tracking technology to the observation mission will make it more effective and easier. An important requirement was that the tracker must be fast enough to meet the real-time requirements while still ensuring accuracy and stability. In addition, observation equipment usually uses compact hardware (such as embedded computers) and high-resolution cameras. In this paper, a Kernel Correlation Filter (KCF) based tracking algorithm is used with a Raspberry Pi 4B to track objects at sea. The experiment results show that the tracker works well and stably, and the tracking speed reaches 20 FPS with 1280×720 pixels of camera resolution.