kavinh07/

ocr_dataset_shamadhan_synth_30k_p5

Built from kavinh07/ocr_dataset_shamadhan_synth_30k_p2 with 3,200 additional confusion-pair training images.

Split Samples

train 167,949

validation 36,985

kavinh07/ocr_dataset_shamadhan_synth_30k_p2 — original synthetic + shamadhan real data synthetic_confusion — 3,200 confusion-pair images (train only)

	Columns




	

Column Type Description

image Image Cropped NID field (RGB)

text string… See the full description on the dataset page: https://huggingface.co/datasets/kavinh07/ocr_dataset_shamadhan_synth_30k_p5.