discussion / Drones  / 11 July 2025

Counting aggregated animals in orthomosaics?

Hello everyone!

Just wanted to share two papers that we published recently on approaches to deal with counting errors when surveying wildlife populations using drone-derived orthomosaics..

In the first paper, we draw attention to some unintended counting errors that may arise specifically when building orthomosaics when animals are moving during the flight, and we discuss some strategies to mitigate these errors and potential solutions for further explorations. 

In the second paper, we develop an approach to estimate the abundance of aggregated wildlife populations from orthomosaic counts, taking into account multiple sources of variations (entries and departures of individuals during the sampling period, availability during flight, double counts..). We apply this approach to estimate the population of Giant South American River Turtles during a mass nesting event, what resulted in the description of the current world's largest known aggregation of freshwater turtles

Hope it is useful and I would be happy to receive any feedback! =)

Ismael

Brack2025 - Counting animals in orthomosaics from aerial imagery_0.pdf Brack2025- Estimating abundance of aggregated populations with drones turtles_0.pdf


Nahuel Emiliano Farias
@nahuel  | he
Instituto de Investigaciones Marinas y Costeras (IIMYC-CONICET/UNMDP)
Argentina-based researcher specializing in the biology and ecology of marine invertebrates, with a focus on decapod crustaceans. Interested in marine monitoring and conservation technology from coastal to deep-sea ecosystems.
Involvement level 1

Great papers, thanks for sharing! 
We’ve been using drones to detect color-tagged crabs in saltmarshes, which allowed us to achieve a 60% recapture rate—something that was impossible with traditional trap or sampling methods.

By comparing orthomosaics captured at consecutive time intervals, we managed to re-detect individuals and quantify their movement for the first time. However, our detection process was done manually by eye.

I’m now looking into automated detection. Do you know of any existing tools that could be adapted for tracking color-coded targets across sequential maps, or would you recommend developing an ad hoc solution?

Best,
Nahuel