Research on object detection and recognition in remote sensing images based on YOLOv11

Area of Science:

• Remote Sensing
• Computer Vision
• Artificial Intelligence

Background:

• Accurate detection of ground objects in high-resolution remote sensing imagery is crucial for various applications.
• Existing methods face challenges in achieving high accuracy and efficiency, especially in complex scenarios.

Purpose of the Study:

• To evaluate the performance of the YOLOv11 model for ground object target detection in high-resolution remote sensing images.
• To assess its potential in improving detection accuracy and efficiency.

Main Methods:

• Training the YOLOv11 model on a dataset of 70,389 samples across 20 target categories.
• Utilizing loss functions (Box_Loss, Cls_Loss, DFL_Loss) and evaluating performance using precision, recall, map50, map50-95, and F1 score.

Main Results:

• The model achieved high performance metrics: precision (0.8861), recall (0.8563), map50 (0.8920), map50-95 (0.8646), and F1 score (0.8709).
• Rapid convergence of loss functions indicated effective optimization.
• 80% of test samples achieved confidence scores above 85%, confirming reliability.

Conclusions:

• YOLOv11 shows significant promise for remote sensing image target detection, offering high accuracy and robustness.
• The model provides robust technical support for intelligent remote sensing image analysis.
• Future work will focus on dataset expansion, model refinement, and performance improvement for small targets and complex scenes.