AI-Enabled Smart Camera Traps for Wildlife Monitoring in African Ecosystems

Abstract:

Wildlife monitoring is essential for conservation, yet traditional camera traps depend on manual data retrieval and offline processing, limiting their effectiveness. This work-in-progress investigates a low-cost, AI-powered smart camera system leveraging YOLO models for real-time animal detection in African ecosystems. We evaluate these models on a curated African wildlife dataset, with YOLOv11 achieving mean average precision (mAP) scores above 0.808, demonstrating their suitability for edge deployment. A Raspberry Pi–based prototype equipped with a Hailo AI accelerator facilitates efficient on-device inference. Preliminary field tests indicate strong model generalization and fast inference speeds, validating the system’s practical applicability. These results show the feasibility of deploying lightweight deep learning models in remote conservation locations, enabling scalable and autonomous biodiversity monitoring with minimal infrastructure requirements.

Authors:

Hadise Pishdast, Hari Kalva & Donovan Tye

Conference / Journal

In Proceedings of the 2025 International Conference on Information Technology for Social Good (pp. 107-112)