CCExtractor/Rekognition

Add Support For API Selection in Text Detection

augsaksham opened this issue · 1 comments

There are two text detection API in the codebase:

text_reco_url = 'models/crnn_syn90k_saved_model:predict'
text_detect_url = 'models/east:predict'

Currently the model : "crnn_syn90k_saved_model" is used for making predictions.

Add a feature to allow user to select between the model they want to use for the recognition.

@augsaksham these 2 models serve different purposes
text_detect_url = 'models/east:predict' is for text detction i.e localizing text in scene images
text_reco_url = 'models/crnn_syn90k_saved_model:predict' is for text recognition i.e reading the localized text