Open Access

ARTICLE

A Lightweight and Optimized YOLO-Lite Model for Camellia oleifera Leaf Disease Recognition

Qiang Peng^1,2, Jia-Yu Yang¹, Xu-Yu Xiang^1,*

1 College of Computer and Mathematics, Central South University of Forestry and Technology, Changsha, 410000, China
2 College of Aeronautical Engineering, Hunan Automotive Engineering Vocational University, Zhuzhou, 412000, China

* Corresponding Author: Xu-Yu Xiang. Email:

(This article belongs to the Special Issue: Advances in Artificial Intelligence for Engineering and Sciences)

Journal on Artificial Intelligence 2025, 7, 437-450. https://doi.org/10.32604/jai.2025.072332

Received 24 August 2025; Accepted 22 September 2025; Issue published 20 October 2025

Abstract

Camellia oleifera is one of the four largest oil tree species in the world, and also an important economic crop in China, which has overwhelming economic benefits. However, Camellia oleifera is invaded by various diseases during its growth process, which leads to yield reduction and profit damage. To address this problem and ensure the healthy growth of Camellia oleifera, the purpose of this study is to apply the lightweight network to the identification and detection of camellia oleifolia leaf disease. The attention mechanism was combined for highlighting the local features and improve the attention of the model to the key areas of Camellia oleifera disease images. To prove the recognition of the optimized network on Camellia oleifera leaf disease, we tested the network performance of the optimized model with other object detection algorithms such as YOLOV5s, SSD, Faster-RCNN, YOLOv8, and YOLOv10. The results show that the mAP, recall, and accuracy of the trained network achieved 82.9%, 75.7% and 80.6%, respectively. The optimized YOLO-Lite model has the advantages of small size and few parameters while ensuring high accuracy, thus it has a satisfactory effect on leaf disease identification of Camellia oleifera.

---

Keywords

---