Fruit Quality Detection: An Automated Approach

India is an agricultural country and ranks second in the production of fruits globally, following China, with different types of fruits cultivated across the nation. The manual procedure is still followed at every stage of agriculture which is time consuming. Automation in fruit quality detection and sorting will always result in reduction of time-to-market, thus increasing profit. In this paper, considering external quality factors, an algorithm for fruit quality detection is developed using Deep Learning technique based on external quality factors. The color and shape parameters are used to determine the quality of fruit using convolution neural network (CNN). Deep learning models are trained on Google Colab and Tensorflow. Using CNN, with 567 instances of various fruits, an accuracy of 99.1 % with precision of 97.3% is obtained. The performance of CNN is compared with support vector machine (SVM) using accuracy, Fl score and precision. A real-time implementation is done using Arduino-UNO along with small-scale mechanical setup.