A synthetic pre-training dataset containing 53,500 images

This ZIP file contains 53,575 pages of synthetic handwritten images. These images are synthesized by simulating real archival image patches. In addition to the PNG-format images, the file also includes the original annotation file in TXT format (synthetic_Data_label_PaddleLabel.txt). This file is directly exported from the PaddleLabel annotation platform. After configuring the cache file (Cache.cach, whose content is identical to that of Label.txt) and the state file (fileState.txt, which indicates whether an image has been fully annotated), the annotations can be viewed and edited directly within the PaddleLabel platform.

A synthetic pre-training dataset containing 53,500 images.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:47ba82c4c5f1b6658a3bfc81380c1bd078e89e33827051bf5ad6afc2fc8ac442
+size 5622296248