Classification of Tomato Maturity Using Transfer Learning EfficientnetB0 Algorithm

In its evolution, the agricultural industry in Indonesia requires advanced technology, particularly in the field of AI. One application is the use of AI in image classification for sorting tomato ripeness. Manual sorting of tomato ripeness proves ineffective and inefficient, prompting the development of an automatic tomato sorting application based on the EfficientNetB0 model. The initial stage involves comparing various CNN models, including VGG-16, VGG-19, ResNet-50, EfficientNetB0, and a custom CNN model. The accuracy results are as follows, in order: 93 %, 94 %, 42<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">%</sup>, 98%, and 48.2%. EfficientNetB0 emerges as the most accurate model from the comparison. Consequently, this model will be deployed as an application to automate the classification of tomato ripeness levels.