Monocular Deep Learning Pipeline for 3D Bird Collision Risk Assessment on Power Transmission Lines
Abstract
Bird collisions with transmission lines represent a major environmental challenge for wildlife conservation. Estimation of bird mortality from collisions remains particularly limited in areas where the site accessibility for carcass surveys is restricted or impractical. This study evaluates whether a single low-cost video camera, combined with state-of-the-art computer-vision techniques, can deliver reliable collision-risk metrics and thus address this shortfall. We introduce the ARA–01 and ARA–02 datasets, the first specifically designed for detecting small, fast-moving birds in complex riverine environments. Three-dimensional flight paths were reconstructed by integrating a YOLOv11-L model (mean Average Precision at IoU = 0.5 [[email protected]] = 0.799; F1-score = 0.71), monocular depth maps from Depth Anything V2, and frame-wise Random Forest regression anchored to spherical markers. This approach achieved a Root Mean Square Error (RMSE) of 22.38 m and coefficient of determination (R2) = 0.929 across the 50–330 m range. Field trials on 10 and 14 October 2024 recorded 32 flights, of which only seven approached within 10 m of the wires. With 95 % confidence, none entered the <5 m high-risk corridor, though the small sample size and limited temporal window preclude conclusions regarding avoidance behavior or deterrent efficacy; further long-term monitoring is required before any causal interpretation. The system therefore provides a practical low-cost monocular setup capable of automatically generating risk indicators.
Related
- From Connection Loss to Collision: Closed-Loop Fault Assessment of a Fly-Connectome SNN
- An automated pipeline for adult and developmental neuropsychological assessment of executive functions and metacognition based on a multi-layer neuropsychological digital-twin framework
- An integrated fuzzy multi-criteria decision making and mixed-integer linear programming optimization framework for project risk assessment and response strategy selection: A case study of a liquefied petroleum gas pipeline construction project
- Sparse Machine Learning Pipeline with Stabl Identifies Cord Blood Multi-Omic Signatures of Bronchopulmonary Dysplasia
- Optimizing non-invasive assessment of gastric activity: a fully automated channel-wise pipeline for multi-electrode electrogastrography
BibTeX
@article{medina2026monocular,
title = {{Monocular Deep Learning Pipeline for 3D Bird Collision Risk Assessment on Power Transmission Lines}},
author = {Pedro Lopes Ismerim Medina and Marco Antonio Ruiz Rueda and Benito Rafael Santana de la Torre and Ícaro Menezes Pinto and Pedro Ricardo Nery Nunes Silva and Gabriela Gama da Silva Santos and Viviane Spencer Andrade and Tacito Henrique da Silva Graça and Luiggi Cavalcanti Pessôa and Alessandra Schwertner Hoffmann and Michele Ferreira Lima and Ricardo Abranches Felix Cardoso Junior and Beatriz Cabral Dias de Carvalho},
journal = {Brazilian Archives of Biology and Technology},
year = {2026},
month = oct,
doi = {10.1590/1678-4324-2026250878},
url = {https://doi.org/10.1590/1678-4324-2026250878},
note = {CC BY},
}
RIS
TY - JOUR
TI - Monocular Deep Learning Pipeline for 3D Bird Collision Risk Assessment on Power Transmission Lines
AU - Pedro Lopes Ismerim Medina
AU - Marco Antonio Ruiz Rueda
AU - Benito Rafael Santana de la Torre
AU - Ícaro Menezes Pinto
AU - Pedro Ricardo Nery Nunes Silva
AU - Gabriela Gama da Silva Santos
AU - Viviane Spencer Andrade
AU - Tacito Henrique da Silva Graça
AU - Luiggi Cavalcanti Pessôa
AU - Alessandra Schwertner Hoffmann
AU - Michele Ferreira Lima
AU - Ricardo Abranches Felix Cardoso Junior
AU - Beatriz Cabral Dias de Carvalho
PY - 2026
DA - 2026/10/27
JO - Brazilian Archives of Biology and Technology
DO - 10.1590/1678-4324-2026250878
UR - https://doi.org/10.1590/1678-4324-2026250878
AB - Bird collisions with transmission lines represent a major environmental challenge for wildlife conservation. Estimation of bird mortality from collisions remains particularly limited in areas where the site accessibility for carcass surveys is restricted or impractical. This study evaluates whether a single low-cost video camera, combined with state-of-the-art computer-vision techniques, can deliver reliable collision-risk metrics and thus address this shortfall. We introduce the ARA–01 and ARA–02 datasets, the first specifically designed for detecting small, fast-moving birds in complex riverine environments. Three-dimensional flight paths were reconstructed by integrating a YOLOv11-L model (mean Average Precision at IoU = 0.5 [[email protected]] = 0.799; F1-score = 0.71), monocular depth maps from Depth Anything V2, and frame-wise Random Forest regression anchored to spherical markers. This approach achieved a Root Mean Square Error (RMSE) of 22.38 m and coefficient of determination (R2) = 0.929 across the 50–330 m range. Field trials on 10 and 14 October 2024 recorded 32 flights, of which only seven approached within 10 m of the wires. With 95 % confidence, none entered the <5 m high-risk corridor, though the small sample size and limited temporal window preclude conclusions regarding avoidance behavior or deterrent efficacy; further long-term monitoring is required before any causal interpretation. The system therefore provides a practical low-cost monocular setup capable of automatically generating risk indicators.
SN - 1678-4324
SN - 1516-8913
ER -