Classification of Ficus Carica Variants using Transfer Learning

Ficus Carica, commonly known as fig, is a fruit with high nutritional value. With numerous variants, each possessing distinct economic significance, confusion arises during the cultivation of its saplings due to the striking similarities among variants. Consequently, losses are often realized long after the trees are planted. This research aims to develop a model capable of distinguishing fig leaf variants. A dataset comprising 400 fig leaf images categorized into four classes, namely Green Jordan, Blue Giant, Banane, and Thailand Brown Turkey, was employed to train the model. The architecture chosen for the model was VGG16, and its performance was evaluated using the 10-cross validation method. Experimental results demonstrated a classification accuracy exceeding 81.50%, indicating the model’s effectiveness in distinguishing fig leaf variants. These findings underscore the suitability of fig classification for practical applications, both at an individual and industrial scale, emphasizing its potential to mitigate losses caused by misidentification during cultivation.