Motion-aware Event Suppression for Event Cameras
This work addresses the problem of filtering unwanted events for event camera users, providing significant improvements in both performance and efficiency for downstream applications.
This paper presents a framework for Motion-aware Event Suppression that filters events caused by Independent Moving Objects (IMOs) and ego-motion in real time. The model achieves 67% higher segmentation accuracy and 53% higher inference rate on the EVIMO benchmark compared to prior state-of-the-art methods, while also accelerating Vision Transformer inference by 83% and reducing Absolute Trajectory Error (ATE) in visual odometry by 13%.
In this work, we introduce the first framework for Motion-aware Event Suppression, which learns to filter events triggered by IMOs and ego-motion in real time. Our model jointly segments IMOs in the current event stream while predicting their future motion, enabling anticipatory suppression of dynamic events before they occur. Our lightweight architecture achieves 173 Hz inference on consumer-grade GPUs with less than 1 GB of memory usage, outperforming previous state-of-the-art methods on the challenging EVIMO benchmark by 67\% in segmentation accuracy while operating at a 53\% higher inference rate. Moreover, we demonstrate significant benefits for downstream applications: our method accelerates Vision Transformer inference by 83\% via token pruning and improves event-based visual odometry accuracy, reducing Absolute Trajectory Error (ATE) by 13\%.