Method and system for intelligent identification and correction of questions

A method for intelligent identification of a question, comprising:

• Step 1:
• identifying regions of individual questions on a test paper based on an image of the test paper and a first model trained in advance, wherein the first model is a neural network based model;
• Step 2:
• identifying question type for regions of the individual questions respectively by using the first model, and labeling the identified individual types of questions respectively, and labeling individual components of the corresponding type of the question in the region of individual questions, the components including stems, answers, and/or pictures;
• and
• Step 3:
• identifying characters in the region of the individual questions based on the separate images of the regions of the individual questions on the test paper, the type of the question corresponding to the region of the question, the component corresponding to the region of the question, and a second model, so as to determine the information recorded on the test paper, wherein the second model is a neural network based model.