Related Experiment Video
Updated: May 11, 2026
03:31
Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications
Published on: December 15, 2023
475
Automated classification in turtles genus Malayemys using ensemble multiview image based on improved YOLOv8 with CNN
Wararat Songpan1, Thotsapol Chaianunporn2, Khemika Lomthaisong3
1Department of Computer Science, College of Computing, Khon Kaen University, Khon Kaen, 40002, Thailand. wararat@kku.ac.th.
Scientific Reports
|October 24, 2024
Summary
Automated image analysis using an improved YOLOv8 model accurately identifies three Thai snail-eating turtle species. This computational biology advancement aids wildlife conservation by enabling rapid, expert-level species identification.
Area of Science:
- Computational Biology
- Wildlife Conservation
- Herpetology
Background:
- Two Malayemes snail-eating turtle species in Thailand are protected, but a newly identified species, M. khoratensis, is not.
- Accurate species identification is crucial for enforcing wildlife protection laws but is challenging due to ambiguous morphological characteristics.
Purpose of the Study:
- To develop an automated, image-based tool for reliably differentiating the three Malayemes turtle species.
- To improve upon existing identification methods by leveraging advanced image processing and machine learning.
Main Methods:
- An ensemble multiview image processing approach was developed, enhancing the YOLOv8 architecture with a convolutional neural network (CNN).
- The model analyzes images from multiple angles to capture unique morphological features, addressing variations and occlusion.
- A comprehensive dataset was used to train and validate the model, comparing its performance against non-ensemble and single-view strategies.
Main Results:
- The proposed ensemble multiview YOLOv8 model achieved a high average mean average precision (mAP) of 98% for Malayemys species classification.
- The accuracy of the image analysis model was validated against genetic methods using mitochondrial DNA, which showed distinct sequences for each species.
- The model effectively overcomes the ambiguity in morphological characteristics for species identification.
Conclusions:
- The developed image analysis tool provides a rapid and accurate method for identifying Malayemes turtle species, supporting conservation efforts.
- This approach advances computational biology and offers a valuable alternative to traditional identification methods, increasing confidence in field identification.
- The study contributes to the enforcement of wildlife protection laws by enabling precise identification of protected turtle species.
