Method, system and storage medium for intelligent scoring of subjective questions based on deep learning

Abstract

The invention discloses an intelligent marking method, system and storage medium for subjective questions based on deep learning, including: acquiring an image of an answer sheet; preprocessing the acquired image; using OpenCV image segmentation processing to segment the image of the answer sheet, It is divided into the answer area of ​​objective questions and the answer area of ​​subjective questions; then, the OMR method is used to identify the answer area of ​​objective questions; the OCR method is used to identify the answer area of ​​subjective questions; the standard answers of objective questions and subjective questions are all entered into the database; the subjective questions include: subjective questions with standard answers and subjective questions without standard answers; the scores of objective questions and subjective questions are counted sequentially. If a test paper with an abnormal score is found during the marking process, manual review intervention is required to correct the deviation of the abnormal test paper.

Description

technical field [0001] The present invention relates to the field of computer-aided examination papers, in particular to a method, system and storage medium for intelligent examination papers of subjective questions based on deep learning. Background technique [0002] In recent years, the answer sheet recognition system has been well-known by the society for many years and has been developing and progressing. With the advent of the big data era and cloud computing, online marking is also gradually improved and perfected according to needs. At present, it mainly includes traditional optical character recognition OCR (Optical Character Recognition) and optical mark recognition OMR (Optical Mark Recognition). Input it into the computer, and carry out an effective recognition process on the information in the image. [0003] OCR (Optical Character Recognition) method: first, scan and record the information to be processed or other documents through the acquisition tool; then p...

AI Technical Summary

Problems solved by technology

At the same time, OMR also has some limitations in practical applications. If the "painted spots" on the information card are not completely aligned with the electric eye, it cannot be accurately identified, that is, it cannot be accurately identified when the information card is tilted; the wrinkled information card cannot be identified; Paper with low printing quality and information cards with low quality paper itself cannot be recognized; the marking must be filled in according to the specifications, otherwise the recognition effect will be greatly affected

Therefore, in the actual application environment, the information card is scanned and imaged by the scanner, and there will be recognition errors when it is tilted

Images