Perception module for Intelligent Parking System deployed on Edge

This repository contains code for a parking edge system designed to detect vehicles using a GoPro camera setup with a Jetson device. Follow the steps below to set up and use the system effectively.

Video Demonstration

For a visual demonstration of this project, please refer to the video linked below:

Project Video Demonstration

Prerequisites

• Ensure the edge system is up to date and running a Linux distribution with Nvidia drivers installed.
• Install the correct version of PyTorch that supports CUDA.

GoPro Camera Setup

1. Follow the instructions provided in the GoPro as Webcam on Linux repository to set up your GoPro camera as a webcam on your Linux system.
2. Ensure that the GoPro camera is detected as USB-connected on the GoPro screen.

Calibration of GoPro Camera

1. Refer to the step-by-step tutorial here to calibrate your GoPro camera for optimal performance with the parking edge system.

Testing Modules

1. Navigate to the "test_modules" folder.

2. Run the following commands in separate terminals to test the camera settings and detection code:

   • Step 1: Define Parking Spots

     python3 set_regions.py
     

     Follow the instructions provided in the terminal to define parking spots in the camera frame.
     
     

   • Step 2: Detection Module

     python3 Detection.py
     

     Ensure to put the MongoDB connection string in the "Detection.py" file to connect to the database.
     
     

   
   • Step 3: Localization Module

     python3 Localization.py
     

     Ensure to put the MongoDB connection string in the "Localization.py" file to connect to the database. Additionally, create an API key on Google Console for localization purposes.