แจ้งเอกสารไม่ครบถ้วน, ไม่ตรงกับชื่อเรื่อง หรือมีข้อผิดพลาดเกี่ยวกับเอกสาร ติดต่อที่นี่ ==>
หากไม่มีอีเมลผู้รับให้กรอก thailis-noc@uni.net.th
หากไม่มีอีเมลผู้รับให้กรอก thailis-noc@uni.net.th
. . (). . : , .
Performance evaluation of YOLO-based models for coral species detection
Name: Bo Bo, Aung Hein
Organization :
Rangsit University. College of Digital Innovation Technology
Email :
aunghein.b67@rsu.ac.th
Name: Karn Yongsiriwit
Organization :
Rangsit University. College of Digital Innovation Technology
Email :
karn.y@rsu.ac.th
keyword:
Object detection
LCSH:
Coral reefs and islands
-- Monitoring
LCSH:
Marine biodiversity
LCSH:
Image processing
-- Digital techniques
LCSH:
Underwater photography
-- Analysis
Abstract:
Coral reefs are vital for marine biodiversity but are increasingly endangered by climate change, pollution, and human activities. Therefore, monitoring these habitats is critical for conservation, yet conventional survey methods are time-consuming and lack stability. This study addresses these difficulties by evaluating the performance of recent YOLO (You Only Look Once) object detection modelsYOLOv8 through YOLOv12for automatic coral species detection in underwater imagery. The project utilizes 646 annotated underwater images containing six coral species from the SCoralDet dataset, originally developed by the Coral Germplasm Conservation and Breeding Center at Hainan Tropical Ocean University, China. The dataset was pre-annotated and preprocessed using Roboflow and used for training and evaluation with the Ultralytics YOLO framework in Python. Models were evaluated based on key performance metrics: mean Average Precision (mAP@50), precision, recall, and inference speed. Among the models, results show that YOLOv9 achieved the highest detection accuracy (mAP@50 = 81.6%), while YOLOv11 produced nearly comparable results (mAP@50 = 80.9%) with more consistent performance across evaluation metrics and faster inference. These findings illustrate the potential of recent YOLO models for precise, real-time detection of coral species and provide a comparative study that can inform future uses of deep learning in the monitoring of marine biodiversity
King Mongkut's University of Technology North Bangkok. Central Library
Address:
BANGKOK
Email:
library@kmutnb.ac.th
Created:
2025
Modified:
2026-03-06
Issued:
2026-03-06
บทความ/Article
application/pdf
BibliograpyCitation :
In Prince of Songkla University, Phuket Campus. College of Computing. The 9th International Conference on Information Technology (InCIT 2025) (pp.605-610). Phuket : Prince of Songkla University
eng
©copyrights King Mongkut's University of Technology North Bangkok
RightsAccess:
ใช้เวลา
0.023274 วินาที
Bo Bo, Aung Hein
| Title | Contributor | Type |
|---|---|---|
| Performance evaluation of YOLO-based models for coral species detection
มหาวิทยาลัยเทคโนโลยีพระจอมเกล้าพระนครเหนือ Bo Bo, Aung Hein ;Karn Yongsiriwit | บทความ/Article |
Karn Yongsiriwit
| Title | Contributor | Type |
|---|---|---|
| Phishing email classification using few-shot learning with a lightweight LLM (Llama 3.2)
มหาวิทยาลัยเทคโนโลยีพระจอมเกล้าพระนครเหนือ Ei, La Min;Karn Yongsiriwit | บทความ/Article | |
| Improving e-commerce product classification via few-shot learning of Gemma2:2B
มหาวิทยาลัยเทคโนโลยีพระจอมเกล้าพระนครเหนือ Thant, Zar Kyi ;Karn Yongsiriwit | บทความ/Article | |
| Performance evaluation of YOLO-based models for coral species detection
มหาวิทยาลัยเทคโนโลยีพระจอมเกล้าพระนครเหนือ Bo Bo, Aung Hein ;Karn Yongsiriwit | บทความ/Article |