Real-time Road Traffic Surveillance using Cloud

Ivy

Ivy is an open source video-based vehicle counting system which employs several computer vision techniques to detect, track and count vehicles in a traffic scene.

Requirements

• Python 3
• AWS account

AWS Services to be setup:

• EC2
• Lambda
• Media Convert
• S3
• Cloudfront

Setup

• Clone this repo
• Install the dependencies in requirements.txt pip install -r requirements.txt.
• Install detector YOLO's weights and place the content file in the data/detectors/yolo directory.

Detector	Description
yolo	Perform detection using models created with the YOLO (You Only Look Once) neural net. https://pjreddie.com/darknet/yolo/

Run(Only for testing ivy model locally)

• Create a .env file (based on .env.example) in the project's root directory and edit as appropriate.
• Run python -m main.

Run(Cloud vehicle counting)

• Create instances mentioned above.
• Replace bucket names with your bucket name.
• Change the path of the local video file.
• Start the server by running python Run.py.
• Start the consumer on the cloud by running python consumer.py.
• Start the producer on the local by running python producer.py.
• Run the codes in Lamda to setup the file conversion and log writting.
• Setup the web application by running the HTML on the web browser.
• Refresh the webapp to see the output.