carImageSearch

An MVC-patterned Python application which works on vectorized image data written to DataStax Astra DB. The application uses the Flask library to build a small front end to allow for searching on the image file data. It ultimately uses the astrapy library to access and work with the underlying vector embeddings in Astra DB.

A running version of this repository's application can be found at: https://aploetz68.pythonanywhere.com/

Requirements

• A vector-enabled Astra DB database
• An Astra DB application token
• An Astra DB API endpoint
• Environment variables defined for: FLASK_ENV, FLASK_APP, ASTRA_DB_APPLICATION_TOKEN, and ASTRA_DB_API_ENDPOINT:

export ASTRA_DB_APPLICATION_TOKEN=AstraCS:GgsdfSnotrealHqKZw:SDGSDDSG6a36d8526BLAHBLAHBLAHc18d40
export ASTRA_DB_API_ENDPOINT=https://b9aff773-also-not-real-d3088ea14425-us-east1.apps.astra.datastax.com
export FLASK_ENV=development
export FLASK_APP=carSearch

You can use a .env file for the vars as well.

• A local static/images/ directory containing JPEGs or PNGs to embed.
• A local static/input_images/ upload directory for JPEGs or PNGs to search by.
• Python dependencies: sentence-transformers, astrapy, flask and flask_wtf. These are found in the requirements.txt file, and can quickly be installed with the following command:

pip install -r requirements.txt

Functionality

Descriptions and examples for each Python file in the project.

carImageLoader.py

• Creates a new collection named car_images using astrapy.
• Cycles through all files in the static/images/ directory.
• Generates embeddings for each image file.
• Stores vector embeddings and metadata in Astra DB.

Note that most of the images in this repo are from the Car Images Dataset found on Kaggle.

Usage:

python3 carImageLoader.py

carSearch.py

• Builds a small web frontend for the application (View).
• Acts as the main program for the application

Usage:

flask run -p 8000

Terminal output:

 * Serving Flask app 'carSearch'
 * Debug mode: off
WARNING: This is a development server. Do not use it in a production deployment. Use a production WSGI server instead.
 * Running on http://127.0.0.1:8000
Press CTRL+C to quit

If you navigate in your browser to http://127.0.0.1:8000, you may search in one of two ways:

• By text string
• By image (click on the "Search" button after selecting the image)

config.py

• Helps configure the environment for Flask.

webforms.py

• A logical way to separate the web forms definitions from the main code.
• Contains the SearchForm.

carServices.py

• Serves the get_car_by_text method (Controller).
• Handles all interactions with the clip-ViT-B-32 sentence transformer.

astraConn.py

• Handles the connectivity with Astra DB (Model).
• Interacts with the vector data using astrapy.