A dataset for the detection of mathematical expressions in camera captured document images acquired in Vietnam
Description
The dataset consists of 6000 Vietnamese camera captured document images that containing mathematical expressions. The dataset is divided into training and testing datasets. The training and testing datasets consist of 5000 and 1000 document images, respectively. The dataset can be used for development and evaluation of the detection of mathematical expressions in camera captured document images issue. This is the first dataset for the development and evaluation of the detection algorithms of mathematical expressions in Vietnamese camera captured document images. The annotation files (in .json format) provide position information of mathematical expressions. Moreover, the dataset provides Latex strings of mathematical expressions that can be applied for the evaluation of recognition algorithms of mathematical expressions. Researchers can use the Intersection of Union (IoU) metric to determine if the detection is correct or not by using the dataset. Please, kindly refer to the following articles when using the dataset: [1] Bui Hai Phong et al., "Mathematical Expression Detection in Camera Captured Document Images", Lecture Notes on Data Engineering and Communications Technologiesthis link is disabled, 2022, 148, pp. 98–109, 2022. [2] Bui Hai Phong et al., "An end-to-end framework for the detection of mathematical expressions in scientific document images", Expert Systemsthis link is disabled, 2022, 39(1), e12800, 2022.
Files
Steps to reproduce
Firstly, we collect scanned document images from Vietnamese users. The documents are scanned by cameras (mobile phones, computer captured images). Users upload the document images using some applications that support students to study mathematics in Vietnam such as: https://apps.apple.com/us/app/dicamon-gi%E1%BA%A3i-to%C3%A1n-l%C3%BD-h%C3%B3a-anh/id1529833740 or https://tuyensinh247.com/. Secondly, we used the Mathpix Snipping tool (https://mathpix.com/) to identify the mathematical expressions in the document images. Both position information and Latex strings of mathematical expressions are provided in the dataset. Thirdly, we manually check and correct the position information and Latex strings of mathematical expressions. Finally, we store the position information and Latex strings of mathematical expressions into .json files. The names of .json files are the same as the names of document images for easy to use.