BSTID.zip

数据集BSTID（全称双语场景文本图像数据集）用于论文“Research on Tamper Detection in Bilingual Scene Images Based on Semantic Conflict”的研究工作<br>--“original”文件夹收纳了未遭篡改及未经过二次编辑的原始训练图片，这些图片均以统一的“yuan_数字标号”命名格式进行标识。--“test”文件夹内的图像是原始图像、篡改图像和二次编辑图像类型图像的混合体，用于性能评估。----若图像以“yuan_数字标号”命名，则代表其为原始图像----若图像以“jie_数字标号”命名，则表明其为二次编辑后的图像，且其数字标号与原始图像的数字标号相对应，便于追溯其原始图像；----而对于仅包含数字标号的图像，则视为篡改图像，同样地，其数字标号可用于查找对应的原始图像。<br>----------------------------------------------------------------------------------------<br>The dataset BSTID (full name: Bilingual Scene Text Image Dataset) is used for the research work in the paper titled "Research on Tamper Detection in Bilingual Scene Images Based on Semantic Conflict."<br>-- The "original" folder contains unaltered and unedited original training images, all identified with a uniform naming format: "yuan_number label."-- The "test" folder contains a mix of original images, tampered images, and secondary edited images, used for performance evaluation.---- If an image is named "yuan_number label," it represents an original image.---- If an image is named "jie_number label," it indicates a secondary edited image, and its number label corresponds to that of the original image, making it easy to trace back to the original.---- For images that only contain a number label, they are considered tampered images, and similarly, their number label can be used to find the corresponding original image.