Two implementation methods of handwritten numbers recognition

Abstract

Abstract In order to realize the recognition of handwritten numbers in daily life by artificial intelligence, to save labor cost and time cost, a machine learning KNN (K-nearest neighbor) algorithm and CNN algorithm (convolution neural network) in deep learning are proposed. According to the calculation, the recognition accuracy of more than 95% can be achieved. In the actual verification, the recognition accuracy of 98.77% and 98.69% are obtained respectively. Practice has proved that these are effective artificial intelligence recognition methods.