Car plates project. FastAPI service (part 3/3)

This is the project for car plate OCR recognition, which include:

1. Neural network segmentation model for car plate area with number selection (part 1/3)
2. Neural network OCR model for plate character recognition (part 2/3)
3. API service for these two models (part 3/3)
4. Additional example how to use API service in Telegram bot

Fast API service is develop according two neural network model for car plate segmentation and OCR.

src/container_task.py contain 2 classes:

• Storage - can save calculated data into *.json (simple "caching" procedure).
• ProcessPlates - can load ONNX models and process car plate numbers. After that, it save results in Storage.

app.py contain FastAPI application, which have 2 handles:

1. get_content - return content according to content_id.
2. process_content - generate content with NN models for uploaded image.

src/container_task.py contain example of calculating procedure.

Used technologies:

• FastAPI
• Dependencies injector containers
• CI/CD (test, deploy, destroy)
• DVC
• Docker
• Unit & Integration tests with coverage report
• Linters (flake8 + wemake)

Disclaimers:

• the project was originally crated and maintained in GitLab local instance, some repo functionality may be unavailable
• the project was created by me and me only

Location for manual test:

• https://car_plates_api.lydata.duckdns.org
• docs https://car_plates_api.lydata.duckdns.org/docs#/default/process_content_process_content_post

Setup of environment

First, create and activate venv: bash python3 -m venv venv . venv/bin/activate

Next, install dependencies: bash make install

Commands

Preparation

• make install - install python dependencies

Run service

• make run_app - run servie. You can define argument APP_PORT

Build docker

• make build - you can define arguments DOCKER_TAG, DOCKER_IMAGE

Static analyse

• make lint - run linters

Tests

• make run_unit_tests - run unit tests
• make run_integration_tests - run integration tests
• make run_all_tests - run all tests