Communication science et technologie
Volume 23, Numéro 2, Pages 1-13
2025-11-18
Accurate Detection System Using Eigen-cam With Yolo Architecture And Drone Images For Rice Panicle Detection
Authors : Belmir Meroua . Difallah Wafa .
Abstract
In rice field phenotyping, accurate identification of rice panicles is an essential process. Nevertheless, the traditional manual method of characterizing rice panicles is labor-intensive and time-consuming. In this study, we explore the use of the YOLOv9 deep learning model for the detection of rice panicle images without human intervention using aerial footage of rice panicles taken by drones. To enhance the interpretability of the model's predictions, we integrated Eigen-CAM, a gradient-based visualization technique that highlights the regions influencing the model’s decision-making process. The activation maps showed that YOLOv9 successfully concentrated on the important features of rice panicles, even with difficulties from thick plants, blockages, and messy backgrounds. Finally, the proposed model achieved outstanding results, with a recall of 74.3%, an mAP50 of 76.8%, and an mAP50-95 of 52.5%.
Keywords
Automated yield estimation ; Rice panicle detection ; YOLOv9 ; drone images ; Eigen-CAM
Les articles similaires
Elfoul Lantri
.
pages 19-38.
Mostefai Messaoud
.
Aitouni Brahim
.
pages 107-124.
Rachid Zaghdoudi
.
Nadir Fargani
.
pages 44-52.
Abdennour Mohamed Amine
.
pages 1458-1471.
بورطال أمينة
.
ص 143-160.
