Klasifikasi Tanaman Hias Monstera Berdasarkan Citra Daun Menggunakan Metode Single Shot Detector – Mobilenet

Abstract

Many sellers, plant cultivators and plant lovers, especially monstera, still find it difficult to identify several types of monstera because there are many types that have similarities between the types of monstera plants. Therefore, many unscrupulous individuals or sellers are not responsible for fraudulent buying and selling of monstera plants by selling cheap monstera types with expensive monstera types. Of course there will be significant losses if the seller and the cultivator and the buyer incorrectly identify the types of monstera. In this study, the Single Shot Detector - MobileNet method was used to classify monstera species. There were 3 types of plants detected, namely Adansonii, Acuminata, and Esqueleto. A total of 1,500 data were used in this study, consisting of 1,200 data as training data and 225 data as validation data. Data testing as much as 75 data taken using a smartphone camera. The use of the SSD-MobileNet method in classifying species on monstera plant leaves produces an accuracy of 96%. These results indicate that the system created using the SSD-MobileNet method has succeeded well in classifying species on monstera plant leaves.