Visual Anomaly Detection on UAVs
| Project Title | UAV Visual Anomaly Detection in the Maritime Environment Using a Neuromorphic Sensor-Processor Platform |
|---|---|
| Project Timeline | Ongoing |
| Researchers | Dr Saeed Afshar, Professor André van Schaik |
| Partners/Collaborators |  |
Project Synopsis
Developing an event-based visual situational awareness and anomaly detection system for UAVs, leveraging advanced sensing and modeling techniques to detect anomalies in real-time.
Project Details
In this project we are developing an event-based visual situational awareness and anomaly detection system for Unmanned Aerial Vehicles (UAVs). Our system aims to detect anomalies in real-time within complex dynamic visual scenes, even in areas with limited signal availability. Leveraging the high-speed, sparse activation, and high dynamic range of event-based sensors, our system can capture superior imaging under various lighting conditions. Additionally, our lab has developed event-based scene mosaicing and anomaly detection methods, which enable us to model the background visual environment encompassing waves, clouds, and fog. By learning to suppress these background features, our system will conduct in-situ scene analysis and anomaly detection in complex dynamic signal environments, reducing power and data requirements significantly and only reacting to unexpected anomalous events.
