Real-Time Event-Based Tracking and Detection for Maritime Environments

TL;DR

This paper presents a real-time event-based tracking system tailored for maritime environments, addressing challenges like wave-induced noise to improve vessel detection and tracking accuracy.

Contribution

It introduces a novel filtering and clustering method that reduces false positives caused by waves, enhancing maritime vessel detection using event cameras.

Findings

Effective filtering reduces wave-induced false positives.

Cluster movement analysis improves vessel tracking accuracy.

System operates in real-time with high reliability.

Abstract

Event cameras are ideal for object tracking applications due to their ability to capture fast-moving objects while mitigating latency and data redundancy. Existing event-based clustering and feature tracking approaches for surveillance and object detection work well in the majority of cases, but fall short in a maritime environment. Our application of maritime vessel detection and tracking requires a process that can identify features and output a confidence score representing the likelihood that the feature was produced by a vessel, which may trigger a subsequent alert or activate a classification system. However, the maritime environment presents unique challenges such as the tendency of waves to produce the majority of events, demanding the majority of computational processing and producing false positive detections. By filtering redundant events and analyzing the movement of each event cluster, we can identify and track vessels while ignoring shorter lived and erratic features such as those produced by waves.