Thai Venomous Snake Identification using Yolo

Conference proceedings article


Authors/Editors


Strategic Research Themes


Publication Details

Author listAreerut Wongmaha and Tuul Triyason

Publication year2024

Start page1

End page6

Number of pages6

URLhttps://wi-iat2024.sit.kmutt.ac.th/register/program-event-2024.html

LanguagesEnglish-United States (EN-US)


Abstract

Thailand's diverse ecosystem harbors numerous snake species, often causing venomous snake bites necessitating antivenom treatment. Accurate identification of snake species enhances treatment success. This study classified 12 Venomous Snake with medical importance in Thailand, using 6,555 annotated images from iNaturalist, Nick Wildlife and the Queen Saovabha Memorial Institute. Images were annotated and split into training, testing and validation sets. YOLOv7, YOLOv8 and YOLOv9 models were employed with batch size 16, image size 640, and epochs 50, 100 and 150. YOLOv9m excelled with Precision 0.85, Recall 0.815, outperforming YOLOv9s and YOLOv9c. The results show that, YOLOv9m proved the most effective for accurate detection, especially in complex environments for venomous snake species differentiation.


Keywords

Computer Visionimage classificationsnake identificationYOLO


Last updated on 2025-25-01 at 00:00