Application of the Mobilenet Model for Pest Detection in Mustard Plants Based on Leaf Imagery

Abstract

Mustard greens (Brassica rapa) are one of the most widely cultivated vegetables in Indonesia due to their high economic value and nutritional content. However, the productivity of mustard plants often decreases because of pest attacks such as armyworms (Spodoptera litura) and diamondback moth caterpillars (Plutella xylostella). The process of pest identification that is still performed manually is considered inefficient and prone to human error. Therefore, a technology-based system is needed to automatically and accurately detect pests. This study aims to develop an Android-based pest detection application for mustard plants by implementing the MobileNet model based on leaf images. The method used in this research is Convolutional Neural Network (CNN) with the MobileNet architecture because it is lightweight and efficient for mobile devices. The dataset used consists of 1,380 mustard leaf images, including 1,241 training data and 139 testing data. The research stages include data collection, image preprocessing, MobileNet model training, and model evaluation using a confusion matrix. The results of this study show that the developed application is capable of detecting the condition of mustard leaves, whether healthy or infected by pests. The MobileNet model achieved a training accuracy of 97% and a validation accuracy of 95%–98%, indicating that the model can effectively recognize leaf damage patterns. In addition, the application was successfully implemented on Android devices with gallery, camera, cropping, and automatic detection features, making it easier for users to identify pests on mustard plants quickly and practically.