Lots and lots of Documents which store an enormous amount of useful data are being left waste. If we are able to process these documents we can get a lot of insights and make predictions on various fields.
What it does
Class Sports News into :Athletics,Football,Cricket,Rugby and Tennis Which is uploaded as a Text file (.txt)
How I built it
Preprocess the data Train a CNN model Use the Model in Flask web app
Challenges I ran into
Unable to save the model in required format.
Loading the model in Web App. Transforming input to Vectors. Varying number of features in input.
Accomplishments that I'm proud of
Trained CNN Model Used Glove Converted keras model to Pickle model. Loaded the Pickle model in web app. Also saved the vector model as pickle. Vectorised the input before loading into model.
What I learned
How to put Deep Learning models into productions
What's next for Document_Classifier
Document Summarisation and Key Fields Retrieval