Event camera method improves tiny drone detection with fewer errors

PointEvent: Rethinking Event-based Tiny Object Detection via Serialized Motion Evidence Accumulation

Computer Vision and Pattern RecognitionArtificial Intelligence

Summary

Detecting tiny drones far away can be hard because special event cameras produce scattered signals that are noisy or incomplete. The authors propose a new method called PointEvent that looks at these signals in a clever order to piece together the motion better. This approach uses less computing power but keeps important details, helping it detect small moving objects more accurately and quickly than previous methods. Their tests show it works well and runs fast.

What this means in practice

  • For drone surveillance teams: Enhance detection of small drones at long distances using event cameras with less computational effort and fewer false alarms.
  • For autonomous vehicle engineers: Improve recognition of small moving objects in noisy event camera data while keeping model size small and inference fast.

Authors

Zongze Wu, Baofeng Jia, Weiqi Yan, Jingyuan Zhang, Yu Zang, Xiaoyu Chen, Jing Han

Abstract

Event cameras offer high temporal resolution and motion sensitivity for tiny UAV detection, yet distant targets generate sparse and fragmented events that are easily overwhelmed by clutter and ego-motion. Existing methods mainly rely on dense event representations or local sparse spatiotemporal modeling, resulting in redundant computation or fragmented modeling of motion continuity across distant asynchronous events. To address this limitation, we introduce serialized motion evidence accumulation, which treats motion continuity as an ordered evidence propagation process. Specifically, the same event stream is organized into locality-preserving spatiotemporal paths and chronology-preserving temporal paths through the latent complementary serializations. Based on this principle, we propose PointEvent, a lightweight event-wise state-space framework that alternates serialized scans across the complementary orders, progressively consolidating fragmented motion evidence beyond fixed local neighborhoods. A high-resolution event branch preserves fine-grained target responses, while compact context modulation suppresses interference. Experiments demonstrate that PointEvent achieves SOTA with the fewest parameters and fastest measured inference among the compared methods. Code: https://github.com/wzz-z/PointEvent