Fruit Quality Assessment using VGG16 and ImageNet Based Transfer Learning Technique

Automatic classification of fresh and rotten fruits plays a substantial role in agriculture as well as food industry. Traditional methods for fruit spoilage detection are manual, inaccurate, time consuming, laborious, and subjective. To circumvent this drawback of conventional techniques, in this work, transfer learning based automated classification of fruit freshness was proposed. VGG16 pre-trained model and ImageNet database was used for creation of the fruit freshness classifier. Categorical accuracy and validation categorical accuracy were calculated to test the efficiency of the model. Data and system hyper parameters were used to increase the accuracy of the model further. Obtained results conclusively establish transfer learning as an accurate tool for fruit freshness classification.