<![CDATA[Agriculture plays a vital role in increasing Gross Domestic Product (GDP), providing employment, contributing to foreign exchange earnings, and supporting environmental conservation. Indonesia has great potential as an agricultural country where population majority relies on agricultural sector for their livelihood. Pamekasan Regency is center of tobacco production development in East Java, with a tobacco plantation area of over 30,000 hectares. However, pest attacks such as caterpillars often damage tobacco plants, reducing productivity and leaf quality. This study implemented AI technology, specifically Convolutional Neural Networks (CNN), to detect caterpillar pests in tobacco plants in Pamekasan. The main focus is on AI development in computer vision using deep learning techniques. The CNN training process involves several stages: convolution, ReLU layers, subsampling/pooling layers, and fully connected layers. The test scenario was conducted by dividing data by 85% training, 10% validation, and 5% testing, as well as tuning parameters for the learning rate and epochs. The model achieved a maximum accuracy of 85% without overfitting at a learning rate of 0.001 and epochs 15. This demonstrates that the CNN deep learning method can effectively identify disease features in tobacco plants. The application of this technology can increase productivity and efficiency in the agricultural sector, supporting a sustainable economy and ecology.Keywords: convolutional neural network, image detection, tobacco pest.]]>
Copyrights © 2025
