Load image dataset with text ground truth

vision
1

I new on pytorch lib and trying to load a dataset i acquire online. Its a OCR in Urdu language. the structure of the dataset is
-Train_image:
-name1.jpg
-name2.jpg

-train_labels.csv
-column 1 (Files name)
-name1
-name2
-column 2 (text in the corrosponding file)
-label1 (urdu text)
-label2 (urdu text)

now how can i load this data to train a CNN and A RNN for detection and recognition?

2

You could write a custom dataset. The pytorch tutorial can be found here.

1 Like