Quality has a direct impact on a customer’s health and willingness. Additionally, it plays a significant role in pricing the banana. Consequently, investigating the method for determining fruit freshness is essential. In this work, we used transfer learning to examine the process by which freshness changes and establish a connection between freshness and storage dates using bananas as an example. The classifier module was used to classify the banana images after the GoogLeNet model was used to automatically extract features from them. This model that is used for the freshness of bananas can determine it with an accuracy of 98.92, which is greater than the human detection. The results indicate that transfer learning provides an automated, accurate, and nondestructive method for determining fruit freshness. In the future, distinguishing vegetables may be enhanced for various applications.