This is a personal project that solves a specific problem that my family had.
My family lives in an apartment building. Sadly, we do not own a reserved parking spot. There are 3 shared parking spots for all the building residents are entitled to park there. In addition, there are additional parking spots on the main street which everyone is entitled to park (some closer, and some farther).
Since we own a car, everytime that we get back home we struggle to find where to park our car. The preferred one is inside the building, otherwise the closest parking slot on the street itself.
We wanted a tool that would let us know in advance whether there is an available parking slot. This insight would help us let us take a decision where to park our car.
This project purpose is to solve the above problem.
The application is:
- Hosted on AWS.
- Completely a serverless solution.
- Code is deployed using terraform.
- Machine learning model is trained using Tensorflow and inference is done using Tensorflow Lite.
The machine learning model - see more information here
The frontend side is an Angular single-page-application.
The backend side using lambda, and it's event-driven. It has 2 parts:
- Client connection handling - using WebSocket API Gateway (connections are persisted to DynamoDB).
- Business logic - while there are connected clients do the following in loop:
- Capture a screenshot from the camera.
- Analyze the screenshot using the ML model.
- Send the results to connected clients.
- Sleep 15 seconds before next update.
The API between the frontend and the backend is using WebSockets:
- The client open a WebSocket connection, and just wait for updates.
- The backend pushes 2 types of messages to the clients: loading event and update event.
- Clone the repository
- Install git hooks
git config --local core.hooksPath .githooks/