Improving Pest Detection in Strawberry Plants Using Transfer Learning Models

Pests pose a significant threat to strawberry production, making early detection crucial for effective control. However, conventional manual detection methods are often time-consuming, inaccurate, and reliant on expert knowledge. To address these limitations, the development of an automated detection system capable of accurately and efficiently identifying pests is necessary. Convolutional Neural Networks (CNNs), combined with transfer learning, have demonstrated great potential in various image recognition tasks, including pest detection in strawberry plants. This study aims to develop an automated pest detection system for strawberry plants using AlexNet based transfer learning. Specifically, this study will evaluate the performance of AlexNet in identifying pests in images of strawberry plants. The AlexNet transfer learning model is trained using a labeled dataset of strawberry plant images, which includes both pest-infested and pest-free samples. The model's performance is then evaluated using standard metrics such as accuracy, precision, recall, and F1-score. Additionally, visualization of the convolutional and pooling layers is employed to analyze and validate the model's effectiveness in extracting relevant features from the images. The experimental results demonstrate that the AlexNet-based transfer learning pest detection system achieves an accuracy of $\mathbf{9 2 \%}$, a recall of $\mathbf{9 1 \%}$, a precision of $\mathbf{9 3 \%}$, and an F1-score of $\mathbf{9 2 \%}$, highlighting its potential as a reliable tool for pest detection in strawberry cultivation.