Remote sensing small target detection optimization based on YOLOv8
Lujianheng1  Zhudandan1  Jiarui1  Wangning2（Corresponding author）

1 School of Artificial Intelligence, Guangzhou Huashang University, Guangzhou, Guangdong, China， 511300；

2 School of Artificial Intelligence, Guangzhou Huashang University, Guangzhou, Guangdong, China， 511300；

Abstract: To enhance YOLOv8's performance in small remote sensing target detection, this study addresses improvements in network structure, feature fusion, label matching, data augmentation, and loss functions. An optimization model is proposed that integrates a lightweight Backbone, a CBAM attention mechanism, and a BiFPN architecture with multi-scale feature fusion, along with a decoupled detection head, to enhance the ability to detect small target features. K-means++ Anchor clustering and a dynamic label matching strategy are combined to improve positioning accuracy and boost training efficiency.

Keywords : remote sensing image; small target detection; YOLOv8; feature fusion; anchor design; loss function optimization

References

[1]Wang H, Yang H, Chen H, et al. A remote sensing image target detection algorithm based on improved YOLOv8[J]. Applied Sciences, 2024, 14(4): 1557.

[2]Zhao X, Zhang W, Xia Y, et al. G-YOLO: A Lightweight Infrared Aerial Remote Sensing Target Detection Model for UAVs Based on YOLOv8[J]. Drones, 2024, 8(9): 495.

[3]Yi H, Liu B, Zhao B, et al. Small object detection algorithm based on improved YOLOv8 for remote sensing[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2023, 17: 1734-1747.

[4]Liu X D, Zhang H, Gong W, et al. YOLO-PDNet: Small Target Recognition Improvement for Remote Sensing Image Based on YOLOv8[C]//2024 International Joint Conference on Neural Networks (IJCNN). IEEE, 2024: 1-9.

[5]Bi J, Li K, Zheng X, et al. SPDC-YOLO: An Efficient Small Target Detection Network Based on Improved YOLOv8 for Drone Aerial Image[J]. Remote Sensing, 2025, 17(4): 685.