IDENTIFICATION OF GUAVA FRUIT MATURITY USING DEEP LEARNING WITH CONVOLUTIONAL NEURAL NETWORK METHOD

摘要

Guava is one of the most popular agricultural commodities. Guava is not only rich in vitamin C but also contains several types of minerals that can counteract various types of degenerative diseases, and maintain body fitness. One type of guava is Red Guava Getas. Identifying the maturity of guava fruit by farmers is still done manually by doing direct visual observations on the fruit to be classified. Weaknesses in performing visual observations are directly influenced by human consistency in the identification process, so that in certain conditions will occur inaccurately. Therefore, a technology is needed to use computer assistance to help identify the results of the examination and conclude the identification results more accurately. This application uses deep learning with the Convolutional Neural Network (CNN) method with LeNet architecture. Making this application uses the Python programming language and Keras as a back-end Tensorflow. From the tests carried out, it is obtained a percentage of 50% for 100 training data and 10 epochs, a percentage of 85% for 100 training data and 20 epochs, a percentage of 92% for 140 training data and 10 epochs, and the last percentage of 100% for 140 training data and 20 epochs.