Handwritten Text - Model Predicts Only " . " for All Images

[likith1908]

I'm using a trained model downloaded from the provided Google Drive link to extract text from images. However, no matter what image I provide, the model always returns " . " as the predicted text, which seems like an issue with either the image preprocessing or the model itself.

Are there any additional preprocessing steps required other than resizing the image to (128, 32)? Could there be an issue with how the model was trained or with the preprocessing pipeline? Should I apply any other transformations to the images?

Thanks!