Automatic correction system of elementary arithmetic images based on deep learning
Mingtao Liu, Xinyu Xia · Journal of Physics Conference Series · 2020
Abstract The elementary arithmetic exercise is corrected by manual traditionally, which is time consuming and inefficient. With the development of machine learning technology, the accuracy of image recognition is gradually improved, which can assist the manual correcting task. In this paper, a convolutional neural network model is built and trained to realize the recognition of numbers and operators in images based on Keras framework. After the training process, an automatic correction system for elementary arithmetic images is developed using the Django framework. The trained model is used to identify and correct the elementary arithmetic images automatically. The user uploads the elementary arithmetic images by the browser without installing any other application software, which can be used to help teachers in practice after some optimizations.