Looking at text-recognition-0012 pre-trained model description page here, the following architecture information was provided:
This is a network for text recognition scenario. It consists of a VGG16-like backbone and bidirectional LSTM encoder-decoder. The network is able to recognize case-insensitive alpha-numeric text (36 unique symbols).
But do you have a paper reference for this architecture?
Thank you for reaching out. The text-recognition-0012 is an Intel pre-trained model, as it is not open-sourced, we cannot give more information than the provided here.
If you have more questions, feel free to ask us.