Closed jvpoulos closed 7 years ago
I'm using this code to train on the Washington Database, which contain binarized and normalized cropped images like this:
which has a ground-truth label of 'williamsburgh'.
With default settings, I do well during training (loss 0.001056, perplexity 1.00) but at test time get a very low character accuracy (11%).
When I visualize on the test set, I'll get blank images returned
even when the translation is correct word.txt
Fixed image loading to work with binary images
https://github.com/jvpoulos/Attention-OCR/commit/0126c782507f2c1d181239431861504848aa02d9
I'm using this code to train on the Washington Database, which contain binarized and normalized cropped images like this:
which has a ground-truth label of 'williamsburgh'.
With default settings, I do well during training (loss 0.001056, perplexity 1.00) but at test time get a very low character accuracy (11%).
When I visualize on the test set, I'll get blank images returned
even when the translation is correct word.txt