Classification of Fruits Using Convolutional Neural Network and Transfer Learning Models
Pathak, R., & Makwana, H.
Automated categorization of freshness of fruits plays pivotal role in the agricultural industry. In conventional method, the grading of fruit is assessed by human being. This method is cumbersome, inconsistent and easily influenced by surrounding. Therefore a fast, accurate and automated system is required for the industrial applications. The current work uses a deep learning based model for classification of fruit freshness. Proposed Convolution Neural Network (CNN) model is implemented by using public dataset named as "fruit fresh and rotten for classification" derived from kaggle. Using the dataset, three varieties of fresh fruits (Apple, Banana, and Oranges) and their rotten category are used for experiment. A deep learning based CNN model is used to extract the characteristics or attributes from the available fruit images. A softmax function then takes the input images and segregates them into fresh and rotten category. Proposed CNN model evaluates the dataset efficiently and gives the accuracy of 98.23%. Results shows that our proposed CNN model is working efficiently in classification of fruits. In current work four Transfer Learning methods are also investigated for classification of fruits. Classification performance comparison proves that Convolutional Neural Network model is more efficient than the transfer learning models.