Precise Spatial Prediction of Rice Seedlings From Large Scale Airborne Remote Sensing Data Using Optimized Li-YOLOv9

• Published in: IEEE journal of selected topics in applied earth observations and remote sensing 2025-01, Vol.18, p.1-13
• Main Authors: Anandakrishnan, Jayakrishnan, Sangaiah, Arun Kumar, Darmawan, Hendri, Son, Nguyen Khanh, Lin, Yi-Bing, Alenazi, Mohammed J. F.
• Format: Article
• Language: English
• Subjects: Accuracy; Agriculture; Airborne remote sensing; Airborne sensing; Aquatic plants; Atmospheric modeling; Autonomous aerial vehicles; Biological system modeling; Computational modeling; Computer architecture; Convolutional neural networks; Crop production; Crops; Deep learning; Edge computing; Food security; Intelligence; Lightweight; Modules; Parameters; precision agriculture; Precision farming; Real time; Remote sensing; Rice; rice seedling detection; Seeding; Seedlings; spatial attention; Spatial data; Spatial distribution; Unmanned aerial vehicles; Weight reduction; YOLO; you only look once (YOLO)
• ISSN: 1939-1404; 2151-1535
• DOI: 10.1109/JSTARS.2024.3505964

Rice is pivotal in global food security and Rice seeding detection in precision agriculture is essential for optimizing crop productivity and efficient resource use. Currently, the spatial distribution and detection of rice seeding are manually done, which is time-consuming. As technology advances, the rapid progress in precision agriculture calls for innovative solutions like lightweight edge-computing vision models for UAV intelligence. However, the limited computational capabilities of UAVs make it challenging to deploy complex object detection models onboard. This paper proposes Li-YOLOv9, an efficient, lightweight, and precise object detection model for detection of rice seedlings from large-scale UAV remote sensing data. The proposed Li-YOLOv9 includes a 3D Feature Adaptation Module (3D-FAM), Convolutional Block Attention Modules (CBAM), Coordinate Attention Module (CAM), and DeepWise Point Convolution (DWC) modules, which offer significant advantages over the original YOLOv9. The compact Li-YOLOv9 with approximately 9 million parameters is significantly lighter than the original YOLOv9 with 60 million parameters, making it ideal for resource-efficient onboard intelligence. Li-YOLOv9 is tested and validated against a real UAV rice seedling dataset. It showcases exceptional detection capabilities, achieving impressive metrics across various performance indicators, including a Mean Average Precision (mAP) of 99.60%, an F1-score of 96.95%, an Average Recall (AR) of 96.70%, and an Average Precision (AP) of 97.20% in comparison to the state-of-the-art (SOTA). Li-YOLOv9 is highly efficient and ideal for UAV deployment for real-time detection of rice seedlings from large-scale remote sensing data.