Abstract
Plants provide and possess many elements that help humans and the whole ecosystem strive and function properly. Plant disease affects food production, the economy, and the health and safety of plants and humans who consume the plants. Identifying the type of plant disease might be a good solution in planning preventive measures against the disease. Rice or Oryza Sativa is one of the predominant foods or crops in the world. Due to rice diseases, the rice production has significantly decreased. The purpose of this study is to develop a system that will identify Tungro, Blast, and Blight which are the most common diseases in the Philippines. These diseases have similar symptoms shown on the leaf which are visible spots, damage, and discoloration on the leaf. The hardware consists of a Pi Camera, LCD, and a microcomputer. The captured image of a damaged leaf was processed using a CNN for diagnosis. The website serves as a GUI where the user can see the output and the previous predictions. CNN can detect if the leaf is healthy and predict if the disease is Tungro, Blast, or Blight. Laplacian function was used to check the blurriness of the image. The method used in this study is Densenet201 with 100 x 100 pixels target size and 3 x 3 convolutional layer. The dataset contains 400 images per prediction, with more than 1,600 images. The system can detect and identify healthy, Tungro, Blast, and Blight with 80 % reliability. Based on the Likert scale and Cronbach’s alpha, the respondents consider the system excellent in detecting and identifying healthy and diseased plants.
