This repository provides a thorough summary of my contributions to Noora Health. Utilizing Google's Gemini ProVision and DocumentAI technologies, I developed strong methods to pull tabular data from various images of paper documents. Furthermore, I led the creation of a RAG model that can effectively answer user questions in various languages, with the help of an advanced translation interface.
The repository Streamlit_OCR is for the OCR part.
- Python version >= 3.9 installed on your local machine
-
Clone the repository to your local machine:
git clone <repository_url> -
Navigate to the directory "Streamlit_OCR":
-
Install all the dependencies:
pip install -r requirements.txt -
Also install Google Cloud CLI installer from here: https://cloud.google.com/sdk/docs/install
-
The Google Cloud CLI is used for authorisation.
-
To know how to use Google Cloud CLI for verification watch this: https://www.youtube.com/watch?v=gpAiUerUdEA
Follow these steps to run the project locally:
- Make sure
gemini_chained.pyis in the same folder asapp_trial.py. - Open
gemini_chained.pyand:
- Enter your API key in line 282.
- Update other details from line 17 to 20 and 263 to 265 as needed.
- In your terminal, run the following command: streamlit run app_trial.py
The repository RAG is for RAG Model using the Q-A bank given to us.
- Final app is in
app.py
For running the code first install all the dependencies by
pip install -r requirements.txt
Then run the app.py file by running the following command on the command propmt to get a stremlit UI
streamlit run app.py
To change the Query please navigate the code in app.py and change the message variable with whatever Query you want to give in.