Keypoint Detection for Measuring Body Size of Giraffes: Enhancing Accuracy and Precision

Giraffes, especially the endangered Masai giraffes, face threats from illegal killing and habitat loss. Accurate body size measurements are crucial for understanding giraffe biology, but their unique body form poses challenges. Previous studies on giraffe size and growth have been limited by small sample sizes, captive animals, or dissected specimens. To address these limitations, this research project presents a solution for giraffe key point detection using photogrammetry and machine learning. The study utilizes a dataset of 29,806 giraffe images annotated with four key points: front bottom hoof, top ossicone, top head, and neck. Through preprocessing techniques and a vision model, accurate keypoints detection is achieved. The proposed method offers an efficient and noninvasive approach to measuring giraffe body size. By contributing to the understanding of giraffe proportions and growth patterns, this research advances the field of giraffe biology. The developed model can aid wildlife conservation organizations in effectively measuring and analyzing giraffe body size in the wild. This work fills a crucial gap in previous research efforts and offers valuable insights for giraffe conservation and management.

Presentation slides
Speaker Details

Solene Debuysere is currently doing a research scientist internship at Microsoft AI For Good Lab in Kenya. Her research project is in partnership with the Wild Nature Institute. In parallel, she is finishing her Masters in Data Science at Institut Polytechnique de Paris. In September, she will start a PhD in the Netherlands on the use of water and electricity maintenance models in the Dutch Caribbean Islands.

Date:
Speakers:
Solene Debusyere
Affiliation:
Microsoft AI for Good Lab, Institut Polytechnique de Paris