Objective: To improve fruit classification accuracy while considering low computational cost and low sensor cost. Methods: A fruit classification system based on capacitive pressure sensor was proposed. The system used support vector machine algorithm with a Gaussian kernel function (GKF-SVM) to classify fruits. The capacitive pressure sensors used were made of thin copper sheets and a layer of vinyl acetate, and these sensors were fixed to the thumb and index finger of a polyamide spandex glove that simulates a robotic hand. The obtained capacitance value was expressed in the form of digital level, and the data was extracted by data processing software, and the capacitance data was processed by SVM algorithm with kernel functions to determine the category of a fruit. Results: The classification results of 11 kinds of fruits in the designed classification system shown that the smart glove using GKF-SVM algorithm could achieve high accuracy classification of fruits, which could trade off between classification accuracy, calculation cost and low sensor cost according to actual fruit classification needs. Conclusion: The research results can be used to develop electronic systems for fruit classification to improve the fruit classification performance of classification manipulators.