<![CDATA[This article discusses the application of smart clustering in the classification of dragon fruit in Banyuwangi Regency. Banyuwangi is the largest production center in Indonesia. However, some farmers experience problems in producing quality fresh fruit that meets market demand. Therefore, a convolutional neural network method is proposed for the dragon fruit classification system to differentiate rotten and fresh fruit. The dragon fruit sample uses 50 fruits which will be trained based on digital images. The results of applying smart clustering show that the best fruit search value achieved an epoch accuracy of 99.24% with 100 iterations. Meanwhile, iterations 50 and 70 get epoch accuracy of 98.88% respectively. The results of fruit classification are used as initial data in determining good (fresh) and bad (rotten) fruit quality. So, dragon fruit farmers can harvest earlier before the disease spreads to other fruit.]]>
