278
we provided estimates of the number of breeding pairs per year to managers and
cross-checked these with estimates from the ground monitoring and colony sampling. As an added value, the spatial coordinates of nests can be used for spatial
analysis and investigate nest aggregation, density and distribution in order to reveal
spatial relationships with environmental factors such as distance to colony edges,
distance to colony centroid, distance to predators, etc.
Keywords UAVs • Slender-billed Gull • Orthomosaic • Bird delineation •
Photointerpretation • Supervised classification • Colony monitoring
Overview
• UAVs are increasingly helpful and accessible for nature conservation and
management;
• Picturing bird colonies with UAVs has proved to be a very efficient procedure for
estimating colony size with low or no disturbances;
• Ground and manned aerial surveys of bird colonies are time-consuming and have
different constraints in retrieving accurate colony size estimates;
• Most of the studies have identified birds and other animals by visual
photo-interpretation;
• Automatic supervised classification provides acceptable results for delineating
ground nesting birds unless under a canopy;
• Some added-values arise from the use of UAV images in bird colony surveys,
such as the spatial distribution of nests, their attributes and digital surface models
of the colony;
This study compares automatic versus manual approaches to delineating lying
and standing gulls on UAV images of a breeding colony at Doñana Natural Space.
Introduction
In this chapter, we evaluate the use of Unmanned Aerial Vehicles (UAVs) for automated mapping of Slender-billed gulls (Chroicocephalus genei) inside the Doñana
Natural Space, southwest Spain. Particular focus was on monitoring bird colony
size (number of breeding birds or lying individuals) and productivity (clutch size,
total number of chicks). Ground validation of colony size estimates was achieved
through concurrent visual surveys.
Unlike previous studies using UAVs to retrieve bird colony information, we used
two different automatic methods to estimate colony size in addition to visual photointerpretation. The selected automatic classification procedures were: (a) Support
Vector Machine algorithm and (b) Random Forests machine learning methodology.
R. Díaz-Delgado et al.
we provided estimates of the number of breeding pairs per year to managers and
cross-checked these with estimates from the ground monitoring and colony sampling. As an added value, the spatial coordinates of nests can be used for spatial
analysis and investigate nest aggregation, density and distribution in order to reveal
spatial relationships with environmental factors such as distance to colony edges,
distance to colony centroid, distance to predators, etc.
Keywords UAVs • Slender-billed Gull • Orthomosaic • Bird delineation •
Photointerpretation • Supervised classification • Colony monitoring
Overview
• UAVs are increasingly helpful and accessible for nature conservation and
management;
• Picturing bird colonies with UAVs has proved to be a very efficient procedure for
estimating colony size with low or no disturbances;
• Ground and manned aerial surveys of bird colonies are time-consuming and have
different constraints in retrieving accurate colony size estimates;
• Most of the studies have identified birds and other animals by visual
photo-interpretation;
• Automatic supervised classification provides acceptable results for delineating
ground nesting birds unless under a canopy;
• Some added-values arise from the use of UAV images in bird colony surveys,
such as the spatial distribution of nests, their attributes and digital surface models
of the colony;
This study compares automatic versus manual approaches to delineating lying
and standing gulls on UAV images of a breeding colony at Doñana Natural Space.
Introduction
In this chapter, we evaluate the use of Unmanned Aerial Vehicles (UAVs) for automated mapping of Slender-billed gulls (Chroicocephalus genei) inside the Doñana
Natural Space, southwest Spain. Particular focus was on monitoring bird colony
size (number of breeding birds or lying individuals) and productivity (clutch size,
total number of chicks). Ground validation of colony size estimates was achieved
through concurrent visual surveys.
Unlike previous studies using UAVs to retrieve bird colony information, we used
two different automatic methods to estimate colony size in addition to visual photointerpretation. The selected automatic classification procedures were: (a) Support
Vector Machine algorithm and (b) Random Forests machine learning methodology.
R. Díaz-Delgado et al.