I am trying to finetune this model on a custom dataset. How would I go about creating an LMDB dataset?

Currently my ground truth file is as follows:

img_filepath1.jpg	label_text1
img_filepath2.jpg	label_text2

I tried using the converters from MMOCR and CDistNet (https://github.com/simplify23/CDistNet) but they do not seem to work. Showed that there were no samples in the datset. I cannot find any converter from gt.txt to train.mdb in the code provided by you.

Please let me know which one you used or share code for the converter you used if possible. Thank you.