ScoreReader: A handwritten score recognition toolkit for examination papers

We developed a ScoreReader, a toolkit that a handwritten score recognition for examination papers based on the region of interest (ROI) filtering. ScoreReader can support automatic calculation of the total score for each examination paper and reduce teachers' workload to help them correct paper...

Full description

Saved in:
Bibliographic Details
Published in2024 27th International Conference on Computer Supported Cooperative Work in Design (CSCWD) pp. 624 - 629
Main Authors Wang, Weigang, Jing, Wei, Cui, Ziyuan
Format Conference Proceeding
LanguageEnglish
Published IEEE 08.05.2024
Subjects
Accuracy
Federated learning
Filtering
Handwriting recognition
Image recognition
Image segmentation
YOLO
Online AccessGet full text

Cover

More Information
Summary:We developed a ScoreReader, a toolkit that a handwritten score recognition for examination papers based on the region of interest (ROI) filtering. ScoreReader can support automatic calculation of the total score for each examination paper and reduce teachers' workload to help them correct papers efficiently. The toolkit exploits the differences in HSV (Hue, Saturation, and Value) between handwritten scores and background to extract candidate regions accurately. In order to avoid heuristic image segmentation, ScoreReader provides a segmentation-free YOLO recognition approach. In this paper, we evaluate the toolkit on an extended mixed dataset of isolated, double-digit, and connected digits using NIST as the base dataset, and the experimental results demonstrate that the toolkit achieves an accuracy of 95.2% for the computation of handwritten scores on the volume, which is satisfactory performance overall.
ISSN:2768-1904
DOI:10.1109/CSCWD61410.2024.10580288