Face Recognition: Primates in the Wild

Abstract: We present a new method of primate face recognition, and evaluate this method on several endangered primates, including golden monkeys, lemurs, and chimpanzees. The three datasets contain a total of 11,637 images of 280 individual primates from 14 species. Primate face recognition performance is evaluated using two existing state-of-the-art open-source systems, (i) FaceNet and (ii) SphereFace, (iii) a lemur face recognition system from literature, and (iv) our new convolutional neural network (CNN) architecture called PrimNet. Three recognition scenarios are considered: verification (1:1 comparison), and both open-set and closed-set identification (1:N search). We demonstrate that PrimNet outperforms all of the other systems in all three scenarios for all primate species tested. Finally, we implement an Android application of this recognition system to assist primate researchers and conservationists in the wild for individual recognition of primates.

• The paper introduces PrimNet, a CNN for automatic face recognition of primates in the wild, specifically designed for conservation using small datasets.
• PrimNet achieved high Rank-1 identification accuracies, reaching up to 93.75% for lemurs and over 75% for golden monkeys and chimpanzees, outperforming other models.
• PrimNet provides a non-invasive tool for conservationists to monitor endangered primates and is offered open-source to foster future advancements.

Face Recognition: Primates in the Wild

The paper "Face Recognition: Primates in the Wild" presents a novel approach to addressing the critical conservation issue of primate individualization. The authors propose a convolutional neural network (CNN) architecture termed PrimNet, specifically designed for automatic facial recognition of primates, including endangered species such as lemurs, golden monkeys, and chimpanzees. The study aims to facilitate non-invasive tracking and monitoring of these primates in their natural habitats, thereby assisting conservation efforts.

Methodology and Datasets

The authors introduce PrimNet, tailored to recognize primate faces, which significantly differ from human faces in terms of facial structure variability. PrimNet incorporates modifications that enhance its suitability for small datasets typical of endangered species. In this context, the study employs datasets encompassing 11,637 images of 280 individual primates, meticulously collected and annotated with key facial landmarks.

The authors conduct meticulous experiments utilizing three different primate datasets—LemurFace, GoldenMonkeyFace, and ChimpFace—to benchmark PrimNet's performance against existing systems. Verification (1:1 comparison) and identification (1:N search) scenarios are thoroughly evaluated, considering both closed-set and open-set conditions. In all test scenarios, PrimNet consistently demonstrates superior accuracy compared to other models, such as FaceNet and SphereFace, achieving notable Rank-1 identification accuracies of up to 93.75%.

Key Findings

• PrimNet achieves a 93.75% Rank-1 identification accuracy for lemurs, significantly outperforming existing systems like LemurFaceID.
• For golden monkeys and chimpanzees, PrimNet scores Rank-1 accuracies of 90.26% and 75.82%, respectively.
• PrimNet's verification accuracy at 1% FAR surpasses other systems, reinforcing its reliability for individual recognition tasks.

Implications

The implications of this research extend beyond mere technical achievement. In practical terms, PrimNet provides an invaluable tool for conservationists and field researchers, enabling efficient non-invasive monitoring of endangered primate species, thereby reducing the reliance on traditional, intrusive tracking methods. This facilitates better management of primate populations, helping mitigate threats such as habitat loss and illegal trafficking.

Future Directions

Looking ahead, further enhancement of PrimNet could involve expanding datasets and incorporating a face detection module to automate landmark annotation tasks, thus improving usability in the field. Moreover, adapting the system for additional primate species may bolster broader conservation efforts.

The authors conclude by offering PrimNet, along with its accompanying Android application, as open-source tools, aiming to foster community-driven advancements in primate face recognition technology. This initiative invites collaboration and further development, potentially leading to greater breakthroughs in biodiversity conservation technology.