idtracker.ai: Tracking all individuals in large collectives of unmarked animals

TL;DR

idtracker.ai is a novel software that accurately tracks and identifies all individuals in large animal groups from video footage, significantly advancing collective behavior studies.

Contribution

It introduces an adaptive deep learning-based algorithm capable of tracking and identifying up to 100 unmarked animals simultaneously.

Findings

High accuracy in individual identification

Effective tracking in large collectives

Adaptive deep networks improve robustness

Abstract

Our understanding of collective animal behavior is limited by our ability to track each of the individuals. We describe an algorithm and software, idtracker.ai, that extracts from video all trajectories with correct identities at a high accuracy for collectives of up to 100 individuals. It uses two deep networks, one detecting when animals touch or cross and another one for animal identification, trained adaptively to conditions and difficulty of the video.