ODD-SEC: Onboard Drone Detection with a Spinning Event Camera
Authors: Kuan Dai, Hongxin Zhang, Sheng Zhong, Yi Zhou
Organizations: Neuromorphic Automation and Intelligence Lab (NAIL) at School of Artificial Intelligence and Robotics, Hunan University, Changsha, China.
The rapid proliferation of drones requires balancing innovation with regulation. To address security and privacy concerns, techniques for drone detection have attracted significant attention.Passive solutions, such as frame camera-based systems, offer versatility and energy efficiency under typical conditions but are fundamentally constrained by their operational principles in scenarios involving fast-moving targets or adverse illumination.Inspired by biological vision, event cameras asynchronously detect per-pixel brightness changes, offering high dynamic range and microsecond-level responsiveness that make them uniquely suited for drone detection in conditions beyond the reach of conventional frame-based cameras.However, the design of most existing event-based solutions assumes a static camera, greatly limiting their applicability to moving carriers--such as quadrupedal robots or unmanned ground vehicles--during field operations.In this paper, we introduce a real-time drone detection system designed for deployment on moving carriers. The system utilizes a spinning event-based camera, providing a 360° horizontal field of view and enabling bearing estimation of detected drones. A key contribution is a novel image-like event representation that operates without motion compensation, coupled with a lightweight neural network architecture for efficient spatiotemporal learning. Implemented on an onboard Jetson Orin NX, the system can operate in real time. Outdoor experimental results validate reliable detection with a mean angular error below 2° under challenging conditions, underscoring its suitability for real-world surveillance applications. We will open-source our complete pipeline to support future research.
Figures & tables
Fig. 1: Illustration of our system operating in an outdoor scenario. Top: The scene with the trajectories of the drone (in red) and our device (in blue), with an inset at the bottom-left showing the rear view of the same scene. Bottom: The detection results from our system. Specific measurement locations are marked with circled letters.
Fig. 2: CAD model of the system. Red and green boxes represent the actuation module and the perception-computing module, respectively.
Fig. 3: Flowchart of the proposed algorithm. The two modules are executed independently. The Pre-processing part receives data from the event camera and converts it into image slices. The Detection part receives and processes the image stream, performs network inference, and calculates the relative bearing to the system based on the detection results.
Fig. 4: Overview of the proposed detection network. Our novel Temporal Fusion Module (highlighted in a green dashed border) enhances detection by analyzing a sequence of frames that capture motion relationships. It employs Feature Channel Gating (FCG) to adaptively weight features, aggregates Temporal Context Vector (TCV) across the segment, and fuses it with features from the detection backbone to improve the network’s target recognition accuracy.
Fig. 5: Coordinate systems and bearing estimation. (a) Image plane: Detection result with a red bounding box marking the drone and a dot indicating its centroid. (b) Spinning platform system: The detected centroid is projected into the 3D coordinate system (origin: camera optical center; y-axis: rotation axis; z-axis: zero-direction). This projection enables bearing calculation.
Fig. 6: Illustration of detection results under varying lighting conditions. Top: High-glare scenario with the sun directly within the camera’s FoV during the afternoon. Bottom: Low-light scenario after sunset.
Illumination
AP
AP 75
AR 100
MDS [pixels]
Standard
62.19
69.70
67.12
14
Low
59.30
63.80
62.60
17
High
55.80
58.20
60.20
-
TABLE II: Evaluation under varying lighting conditions [%]. Minimum Detectable Size (MDS) denotes the smallest target width identifiable by the network under the corresponding scenario. In high-glare scenarios, we deem that size-based comparison is not particularly meaningful; therefore, this aspect is not included in the experimental evaluation.
Metric
w/o Temporal Fusion
w/ Temporal Fusion
AP ↑ [%]
62.19
64.92 ( +2.73)
AP 75 ↑ [%]
69.70
72.80 ( +3.10)
AR 100 ↑ [%]
67.12
69.87 ( +2.75)
AP S ↑ [%]
56.30
59.90 ( +3.60)
Params ↓ [M]
8.94
9.53
TABLE III: Ablation study of temporal fusion. Up or down arrows ( ↑ or ↓ ) indicate whether a higher or lower value is better. Network parameters (Params) represent the total number of learnable weights in the network.
Fig. 7: Illustration of the detection results. Left: Trajectories of the carrier and drone, with four representative time instants marked by circled letters and connected by purple dashed lines. Top right: Corresponding camera images showing the detection results, where drone detection results are highlighted. Bottom right: Comparison of the estimated angles (Est.) with the ground truth (GT), with GPS data interpolated using cubic splines to obtain the drone’s position at the exact detection moments.
Sequences
Mean error[°]
Median error[°]
Max error[°]
static_1
2.063
2.096
3.103
static_2
1.772
1.764
2.611
static_3
1.736
1.715
2.835
static_4
1.858
1.848
3.144
moving_1
2.004
1.857
4.340
moving_2
1.927
1.921
4.308
TABLE IV: Evaluation on self-recorded sequences using angular error [°].
Node
Function
Laptop
Jetson NX
Pre-processing
Data processing
0.969
1.938
Image generation
10.841
30.514
Detection
Input construction
3.762
5.223
Inference
11.440
20.425
Bearing estimation
0.0475
0.0823
Others
15.249
18.467
TABLE V: Computational performance on different platforms [ Time : ms].
Detecting fast-moving objects, such as unmanned aerial vehicle (UAV), from event camera data is challenging due to the sparse, asynchronous nature of the input. Traditional Discrete Fourier Transforms (DFT) are effective at identifying periodic signals, such as spinning rotors, but they assume uniformly sampled data, which event cameras do not provide. We propose a novel per-pixel temporal analysis framework using the Non-uniform Discrete Fourier Transform (NDFT), which we call Drone Detection via Harmonic Fingerprinting (DDHF). Our method uses purely analytical techniques that identify the frequency signature of drone rotors, as characterized by frequency combs in their power spectra, enabling a tunable and generalizable algorithm that achieves accurate real-time localization of UAV. We compare against a YOLO detector under equivalent conditions, demonstrating improvement in accuracy and latency across a difficult array of drone speeds, distances, and scenarios. DDHF achieves an average localization F1 score of 90.89% and average latency of 2.39ms per frame, while YOLO achieves an F1 score of 66.74% and requires 12.40ms per frame. Through utilization of purely analytic techniques, DDHF is quickly tuned on small data, easily interpretable, and achieves competitive accuracies and latencies to deep learning alternatives.
Event cameras excel at detecting small, fast drones, but today's detectors give away their key advantage: they convert the sparse event stream into dense grids and pay dense-processing cost on inputs that are almost entirely empty. We present SparseVoxelDet, to our knowledge the first coordinate-sparse 3D event voxel bounding-box detector: backbone, feature pyramid, temporal reduction, and detection head all operate directly on coordinate-indexed features, with no dense spatial grid at any stage. Building it exposed a hidden failure mode we name support inflation: an input filling a median 0.0652% of the voxel lattice inflates stage by stage until standard pyramid fusion leaves the detection head locally near-dense. We answer with two ideas. Expansion-free inverse-convolution fusion provably creates no active sites beyond the stored backbone supports, cutting head occupancy from a median 78.88% to 10.53%; quality-aligned supervision then recovers more accuracy than preserving sparsity costs. The payoff is measured, not assumed: executing the same trained network densely costs a median 27.5x the work and 4.65x the latency across 5,000 paired frames, with no frame cheaper dense, while the 6.22M-parameter detector reaches 87.01 AP50 on the FRED drone benchmark, ahead of matched dense controls, and holds its lead on the held-out test partition evaluated once. Sparsity, preserved by construction and supervised well, delivers the efficiency and the accuracy together.
Mohamad Yazan Sadoun, Sarah Sharif, Yaser Mike Banad
Inquire Lab, University of Oklahoma, Norman, OK, USA.
In recent years, Unmanned Aerial Vehicles (UAVs) or drones have gained rapid response in terms of security, search and rescue (SAR), border surveillance, etc. Existing monitoring frameworks often struggle to maintain detection consistency when targets undergo significant scale variations due to altitude changes, leading to critical information gaps. To address this issue, this work proposes an integrated real-time detection pipeline for detecting targets through the wireless live drone video feed. Build upon YOLOv8-nano architecture, extensive flight experiments were conducted to determine the detection performance across multiple flight altitudes. Trained on VisDrone2019 dataset, the results of YOLOv8-nano model achieves 57.4%, 41%, 44.8% and 20.3% in precision, recall, mAP and mAP50:95 respectively. While demonstrating on real environment, this analysis revealed that the algorithm achieves near-total detection reliability at altitudes between 16 and 25 meters with the detection frame rate consistently maintained above 41 FPS and reaching a peak of 50 FPS. However, the goal of this work is to enable real-time person detection from an aerial platform via wireless transmission. This approach effectively addresses the dual challenges of identifying targets at varying scales and ensuring near-to-accurate localization during aerial observation.
Centre for Drone Technology Indian Institute of Technology Guwahati Guwahati, India · Dept. of CSE KIIT University Bhubaneswar, India · Drone LAB National Institute of Electronics and Information Technology Bhubaneswar, India