CNN-Based Deep Learning Utilization Model for Identification of Crystal Guava Leaf Diseases

Identification of plant diseases is a crucial step in maintaining plant health and preventing economic losses due to decreased productivity. This research aims to develop an intelligent system capable of identifying diseases in Crystal Guava (Psidium guajava L.) plants using a Convolutional Neural Network (CNN) method. The method combines digital image processing techniques with machine learning to classify Crystal Guava leaf images into two categories: healthy and diseased. The implemented CNN architecture is based on the Xception model, known for its superior performance in image classification tasks. The dataset used consisted of 1,500 Crystal Guava leaf images, including both healthy and diseased leaves. Test results showed that the developed system achieved 94% accuracy in identifying diseases in Crystal Guava plants, surpassing the performance of other architectures such as VGG16 and InceptionV3. This high accuracy demonstrates the model's ability to recognize complex features in Crystal Guava leaf images. These findings contribute to the development of artificial intelligence-based diagnostic tools for early detection of plant diseases. The proposed system is expected to assist farmers, agricultural researchers, and policymakers in making informed decisions to improve the productivity and health of Crystal Guava plants.

Keywords: Crystal Guava, CNN, Xception, image processing, plant disease detection