<![CDATA[The health of Bornean orangutans is increasingly threatened by intestinal nematode parasites, which cause significant morbidity and mortality. Traditional microscopic diagnosis is accurate but slow, labor-intensive, and impractical in remote conservation areas. This paper presents a proof-of-concept smart diagnostic automated system that integrates internet of things (IoT) enabled mobile microscopy with a deep learning model based on you only look once version 11 (YOLOv11). A publicly available dataset of 4,000 annotated parasite egg images, derived from human fecal samples and used as a proxy for orangutan infections, was employed for model training and evaluation. The proposed system achieved a mean average precision (mAP) of 0.9957 and a mean intersection over union (IoU) of 0.9098 across four target classes. Compared with prior works using YOLOv4, YOLOv5, and lightweight models, our approach provides higher segmentation fidelity and is embedded in an IoT-based framework suitable for field deployment. Importantly, a pilot test conducted in the field using real orangutan fecal samples confirmed the system feasibility, with near real-time inference (~300 ms per image) and usability by non-specialist users under low-resource conditions. While broader validation with larger orangutan specific datasets remains necessary, this study demonstrates how IoT and computer vision can be combined into a scalable diagnostic tool for wildlife health monitoring and conservation applications.]]>
Copyrights © 2026
