Banana Variety Classification Using Transfer Learning: Leveraging Pre-trained Models for Accurate Identification

Abstract

In recent times, there have been different varieties of bananas in our country, which we cannot understand when we see them easily. When we go to a shop or market, we cannot tell which type of banana it is, which is a big challenge for us. If we do not know the banana variety properly, we cannot go anywhere and tell anyone that we need this type of banana. Now, when you go to buy bananas but you don't know the types of bananas, you want one kind of banana, but they give you another kind of banana, and you can't understand something. That's why we have a big problem. So, we have determined the different varieties of bananas through the machine, which makes it very easy for us to recognize the varieties of bananas. To solve this problem, we have tried to solve it by using some algorithms through machine learning. Implementing different machine learning techniques, such as pre-trained models, this comparative study analyzes the performance of different machine learning algorithms. The trial results verify that the ResNet50 model has achieved the highest accuracy of 98.09% compared to other VGG16, Xception, and InceptionV3 models.