Event-Based Person Detection for Maritime UAV Search
Organizations: EAVISE-PSI, ESAT, KU Leuven Leuven, Belgium
Abstract
Maritime search and rescue requires locating people in the water quickly, motivating the use of UAVs to survey the search area. During flight, camera motion and the longer exposures used in low light can blur small targets and hinder person detection. Event cameras record brightness changes as they occur, reducing the motion blur associated with longer exposure times. We investigate their use for maritime person detection by generating synthetic events using the v2e event simulator from public MOBDrone rescue videos and retaining the existing annotations. We compare the recurrent vision transformer model RVT-T with YOLOv8n and RT-DETR-R18 trained on exposure-matched images. On 843 test samples from held-out recordings, RVT-T achieves 15.63% person AP50:95, compared with 6.89% for YOLOv8n and 9.44% for RT-DETR-R18 under 100 ms synthetic exposure. Increasing exposure times from 50 to 200 ms progressively reduces frame-based detection accuracy, widening the gap to the event model. To evaluate the feasibility of on-board processing, we deploy our detector on a Jetson Orin Nano platform. The RVT-T model retains its accuracy and requires approximately 136 ms and 2.02 J per sample, allowing real-time processing on the UAV.
Figures & tables
| Condition | Detector | AP 50 | AP 50:95 | Recall | AP S |
|---|---|---|---|---|---|
| Original | RVT-T | 37.42 | 15.63 | 37.25 | 4.59 |
| Original | YOLOv8n | 32.42 | 11.73 | 32.34 | 3.05 |
| Original | RT-DETR-R18 | 35.36 | 15.64 | 37.96 | 3.53 |
| Brightness | RVT-T | 28.91 | 13.65 | 30.18 | 1.43 |
| Brightness | YOLOv8n | 27.36 | 11.08 | 27.31 | 1.31 |
| Brightness | RT-DETR-R18 | 35.81 | 15.29 | 36.41 | 3.92 |
| Detector | Precision | Recall |
|---|---|---|
| RVT-T | 48.67 | 37.25 |
| YOLOv8n | 51.10 | 22.16 |
| RT-DETR-R18 | 31.47 | 30.42 |
| Person AP 50:95 (%) | FP16 processing cost | |||||
|---|---|---|---|---|---|---|
| Detector | FP32 | FP16 | p50 (ms) | p95 (ms) | Power (W) | Energy (J/sample) |
| RVT-T | 15.63 | 15.63 | 135.3–136.8 | 143.2–144.3 | 14.73–14.77 | 2.016–2.024 |
| YOLOv8n | 6.89 | 6.75 | 20.8–21.5 | 21.4–21.7 | 10.84–10.89 | 0.225–0.230 |
| RT-DETR-R18 | 9.44 | 9.50 | 58.6–59.9 | 59.0–60.3 | 12.82–12.95 | 0.762–0.772 |
Object Detection
Object detection locates and classifies objects in images or video, balancing localization accuracy, robustness, and inference speed.
See all 673 papers