Hi folks,
Long time no see :D I hope y'all are doing well. I am writing to discuss a potential sample mobile application idea to demonstrate the usage of text detectors. Currently, we have got two text detectors in TFLite -
I am specifically excited about the CRAFT one because it would show how to take a (complex) PyTorch model and convert it to TensorFlow Lite via ONNX (TFLite team, take NOTE :D). We also have two blog posts in the works -
- End to end tutorial to show the conversion workflow
- Comparative study between EAST and CRAFT to guide the developers to decide which one should be used in different scenarios
Let me know if anyone of you would be interested to build a sample mobile application.