Dry Fruit Classification Using Deep Convolutional Neural Network Trained with Transfer Learning

Dry fruits are an important part of daily life and are most consumable in all parts of the Indian continent. It is difficult to identify the quality of dry fruit with the naked eye. Industrialists or consumers consume quality dry fruits that can be evaluated based on their features. Here in this research, a transfer learning-based deep learning model is proposed to classify multiple dry fruits based on the features such as size, shape, and edge. In this study, transfer learning models such as Xception, Inception, VGG16, VGG19 and SqueezNet are used to set weights for the proposed model and the model is trained with these weights and the dry-fruits dataset. The most appropriate features of dry fruits have been extracted using a fully connected network and combined with transfer learning weight to classify dry fruits. The proposed model was trained and tested with dry fruit images taken from a dataset containing four classes of dry fruits. The proposed model attained a classification accuracy of 98.53%. The model is compared with other deep learning models and the proposed model outperforms.