Format of the training and validation data that is used for training. about ubdvss HOT 2 CLOSED

Hi, thanks for your interest in our work

Our training data has 2-folder structure: "Image" with image data and "Markup" with xml markup data. You can find examples here These examples are public datasets which we publish in our other work.

One way is to structure your data in the same format as we have. When you just run python train.py -t TRAIN_PATH -v VALID_PATH

Alternatively, you may write data reader for your data structure similarly to XMLBarcodeMarkupReader, then specify option -mt YOUR_MARKUP_TYPE when running train.py

from ubdvss.

Hi asmekal. Thank for the response. I am planning to pursue the first path.

from ubdvss.