Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Acoustic Imaging for UAV Detection: Dense Beamformed Energy Maps and U-Net SELD

Belman Jahir Rodriguez, Sergio F. Chevtchenko, Marcelo Herrera Martinez, Yeshwanth Bethi, Saeed Afshar

Abstract

We introduce a U-net model for 360° acoustic source localization formulated as a spherical semantic segmentation task. Rather than regressing discrete direction-of-arrival (DoA) angles, our model segments beamformed audio maps (azimuth & elevation) into regions of active sound presence. Using delay-and-sum (DAS) beamforming on a custom 24-microphone array, we generate signals aligned with drone GPS telemetry to create binary supervision masks. A modified U-Net, trained on frequency-domain representations of these maps, learns to identify spatially distributed source regions while addressing class imbalance via the Tversky loss. Because the network operates on beamformed energy maps, the approach is inherently array-independent and can adapt to different microphone configurations and can be transferred to different microphone configurations with minimal adaptation. The segmentation outputs are post-processed by computing centroids over activated regions, enabling robust DoA estimates. Our dataset includes real-world open-field recordings of a DJI Air 3 drone, synchronized with 360° video and flight logs across multiple dates and locations. Experimental results show that U-net generalizes across environments, providing improved angular precision, offering a new paradigm for dense spatial audio understanding beyond traditional Sound Source Localization (SSL). We additionally validate the same beamforming-plus-segmentation formulation on the DCASE 2019 TAU Spatial Sound Events benchmark, showing that the approach generalizes beyond drone acoustics to multiclass Sound Event Localization and Detection (SELD) scenarios.

Acoustic Imaging for UAV Detection: Dense Beamformed Energy Maps and U-Net SELD

Abstract

We introduce a U-net model for 360° acoustic source localization formulated as a spherical semantic segmentation task. Rather than regressing discrete direction-of-arrival (DoA) angles, our model segments beamformed audio maps (azimuth & elevation) into regions of active sound presence. Using delay-and-sum (DAS) beamforming on a custom 24-microphone array, we generate signals aligned with drone GPS telemetry to create binary supervision masks. A modified U-Net, trained on frequency-domain representations of these maps, learns to identify spatially distributed source regions while addressing class imbalance via the Tversky loss. Because the network operates on beamformed energy maps, the approach is inherently array-independent and can adapt to different microphone configurations and can be transferred to different microphone configurations with minimal adaptation. The segmentation outputs are post-processed by computing centroids over activated regions, enabling robust DoA estimates. Our dataset includes real-world open-field recordings of a DJI Air 3 drone, synchronized with 360° video and flight logs across multiple dates and locations. Experimental results show that U-net generalizes across environments, providing improved angular precision, offering a new paradigm for dense spatial audio understanding beyond traditional Sound Source Localization (SSL). We additionally validate the same beamforming-plus-segmentation formulation on the DCASE 2019 TAU Spatial Sound Events benchmark, showing that the approach generalizes beyond drone acoustics to multiclass Sound Event Localization and Detection (SELD) scenarios.

Paper Structure

This paper contains 22 sections, 4 equations, 8 figures, 1 table.

Table of Contents

  1. Introduction
  2. Related Work
  3. Classical Sound‑Source Localisation (SSL)
  4. Deep Learning (DL) for SSL
  5. Gap Analysis and Motivation
  6. METHODOLOGY
  7. System Overview
  8. Microphone Array Design and Recording Setup
  9. Field Recordings and Data Acquisition
  10. Beamforming Map Construction and Spectral Feature Extraction
  11. Dataset Construction and GPS-Based Labeling
  12. Dataset Representation
  13. U-Net Architecture and Hyperparameter Optimization
  14. Inference and Evaluation
  15. RESULTS
  16. ...and 7 more sections

Figures (8)

  • Figure 1: The overall End-to-end pipeline: multichannel acquisition $\rightarrow$ DAS beamforming over azimuth/elevation $\rightarrow$ spectral binning $\rightarrow$ polar reprojection $\rightarrow$ U-Net segmentation $\rightarrow$ centroid-based DoA estimation.
  • Figure 2: 24 channels microphone array
  • Figure 3: Instance comparison at 101 m between beamforming localization and U-Net inference.
  • Figure 4: False Negative Rate (FNR) across distance bins for Test 2 and Test 1 datasets.
  • Figure 5: Mean angular error across distance bins for Test 2 and Test 1 datasets.
  • ...and 3 more figures