Monkey Tracking - Computer Vision for Automated Macaque Identification and Tracking in the Lab

Supervisors: Samy Rima, Simon Ruffieux, Schmid Michael, Denis Lalanne

Contact person: Samy Rima

Student: Vithusan Ramalingam

Project status: Finished

Year: 2026

This project aimed to develop an integrated system for automated macaque identification and tracking in the lab of the University of Fribourg, building on recent advances in detection, pose estimation, face recognition, and multi-frame tracking.

This Master's thesis presents 3Dmacaque, a markerless system for real-time 3D tracking and behavioral analysis of pairs of macaques using only two conventional cameras. The system combines animal detection, 2D pose estimation, individual identification, and multi-view reconstruction to generate 3D skeletal representations of both animals from either live or previously recorded footage. The proposed pipeline achieves real-time performance at 47.5 frames per second and demonstrates high accuracy in animal detection, keypoint estimation, and individual identification, while substantially outperforming an existing dense-camera approach in processing speed. The system is integrated into a desktop application that supports camera calibration, video analysis, and visualization of behavioral measures such as spatial occupancy, proximity, and movement. An evaluation with four primate researchers showed high usability and perceived usefulness. Overall, the thesis demonstrates the feasibility of accessible, real-time markerless 3D tracking for socially housed macaques using relatively simple hardware, while identifying limited camera coverage and resulting occlusions as the main limitations of the proposed approach.

Interesting references:

Keywords: Machine Learning, Deep Learning, Computer Vision, Dashboard, Data Analysis.

Document: Not yet available