Convolutional Neural Networks Implementation for Chili Classification

Abstract

Horticultural commodities are potential commodities that have high economic value and potential to continue to be developed. One type of potential horticultural commodity to be developed is red chili commodity, especially the red chili. The economy of red chili is quite stable in supplying production to the market. In the process of sorting chili by various processing industry companies, chili exporters and farmers who plant red curly chili is generally done manually and involving humans as the chili decision maker is eligible to be elected or not. The process of identifying manually has many disadvantages, some of which are relatively long time required, humans also tend to feel tired and saturated when doing a monotonous activity, differences in perception of quality, product variety is also obtained because of the human visual limitations, as well as strongly influenced by the psychic condition of the observer. The development of science and digital image processing technology makes it possible to sort the agricultural products and plantations automatically. One of Deep Learning methods technique that is Convolutional Neural Network (CNN) method which currently has the most significant result in image recognition is CNN. The classification accuracy value obtained from the training data is 97.14% and the test data is 80% using RGB input image.