Lightweight 3D U-Net detects bats from ultra-sparse LiDAR data
Needles in a Raystack: Ultra-Sparse LiDAR Occupancy Detection for Bat Tracks
Computer Vision and Pattern Recognition
Summary
Tracking bats at night is hard because their movements create very sparse 3D data when using LiDAR sensors. The authors tackled this by treating the problem as detecting whether tiny cubes in the data are occupied by bats at given times. They designed a small neural network that keeps detailed timing and space information and uses special loss functions to handle the fact that bat signals are very rare compared to empty space. Their method successfully found bat flight paths where traditional approaches failed, making it easier to monitor bats and potentially protect their habitats.
What this means in practice
- •For wildlife conservation teams: Detect bat flying paths from sparse LiDAR scans to monitor nocturnal animal activity over open fields.
- •For wind turbine operators: Integrate bat occupancy detection to improve decisions on turbine curtailment and reduce harm to bats during low-visibility periods.
Authors
Nico Klar, Pankaj Rana, Nizam Gifary, Jakob Traub, Aamir Ahmad
Abstract
Monitoring flying animals is important for understanding and protecting biodiversity, but nocturnal species such as bats are difficult to observe in the field. Using LiDAR, bat movements at night result in ultra-sparse 3D spatio-temporal data in which standard reconstruction losses tend to predict only background and miss real flight paths. We study this problem as voxel-wise occupancy detection in sensor-centric LiDAR raystacks. A lightweight 3D U-Net is proposed that preserves temporal resolution, uses skip connections for spatial detail, and combines weighted binary cross-entropy with Dice loss to handle the strong class imbalance. In real LiDAR recordings of bats over open fields, cross-checked with acoustic monitoring, a reconstruction-based 3D convolutional autoencoder baseline fails to recover foreground trajectories. In contrast, the proposed U-Net recovers sparse foreground occupancy in diagnostic experiments and produces coherent occupancy patterns along bat flight trajectories, providing a practical basis for validation-scale experiments, later clustering of flight tracks, and future integration of bat activity information into biodiversity-aware turbine curtailment strategies.