Skip to content

Latest commit

 

History

History
79 lines (70 loc) · 3.51 KB

README.md

File metadata and controls

79 lines (70 loc) · 3.51 KB

City Assist: Streamlined Platform for Household Help

City Assist is a web application designed to simplify the search for reliable household help by delivering the top 10 relevant matches based on user needs. The platform focuses on efficiency, verification, and user experience, using advanced techniques to ensure accurate and secure results.

Current Working:

Top 5 similar search results of the query “drivers in delhi”.

image

Aadhaar Verification along with Aadhar Card.

image

Nursing Verification using Selenium.

image

Aadhar Card from: https://www.scribd.com/document/432565930/Adhar-card-sample

Methodology

Identity Verification with Selenium

City Assist automates professional ID verification using Selenium for browser interactions, which verifies IDs on official portals. This helps ensure authenticity at scale, especially for specific professions such as nursing.

Aadhaar Verification with PaddleOCR

The platform uses PaddleOCR for Aadhaar card verification. By extracting and matching text from Aadhaar images, City Assist validates user-provided details, enhancing security and reducing manual effort.

Advanced Search with MongoDB Vector Search and Google Embeddings

City Assist employs vector-based search with MongoDB and Google Embeddings, enabling efficient and nuanced matching of user queries with household help profiles. This provides more accurate results compared to basic conditional filtering.

Frontend Development with Streamlit

Streamlit provides a fast, responsive UI for City Assist, enabling users to quickly search, filter, and verify profiles with minimal lag, enhancing the overall experience.

Implementation

City Assist combines vector search with MongoDB, OCR verification with PaddleOCR, and Selenium automation, creating a user-friendly platform for reliable and fast household help matching.

Pre-Requisites:

An account on MongoDB, with network access as 0.0.0.0/0 (access anywhere) or any specific IP address. and preferably a database access with a new user and password to your cluster (for Mongo URI).

  1. Connect to mongodb cluster using, this is MONGO_URI: mongodb+srv://<user_id>:<user_password>@cluster0.ozbxk.mongodb.net/?retryWrites=true&w=majority&appName=Cluster0
  2. Gemini API Key this is GOOGLE_API_KEY: Access it here: https://aistudio.google.com/app/apikey
  3. Chromium driver
  4. Edit MONGO_URI and GOOGLE_API_KEY

Running the Code

  1. Necessary packages:
pip install -r requirements.txt
  1. Load dataset in your own Mongo Database Cluster:
py load.py
  1. Create a Atlas Search Index on Mongo DB, below is the index overview we have used as this has been made using Google Gemini embeddings and Gemini. You can copy paste this on JSON editor portion, after creating search index:
   {
  "mappings": {
    "dynamic": true,
    "fields": {
      "embedding": {
        "dimensions": 768,
        "similarity": "cosine",
        "type": "knnVector"
      }
    }
  }
}
  1. Run the flask server for Aadhaar Verification.
py aadhaarchecker.py
  1. Run flask server for NUID Verification of nurses.
py nursechecker.py
  1. Househelp Data Entry and Verification
streamlit run updated_househelp_ui.py
  1. Househelp Search
streamlit run cityassist.py