Use deep learning techniques to detect and count the number of cars in a parking lot and deploy the same on Amazon EC2 server using FastAPI
In this talk, we will teach how to use different AI algorithms such as OpenCV & Tensorflow to detect and count vehicles in the video streams. Not only we will explain the mechanics behind the AI algorithm but we will also help deploy the same on the cloud server.
A lot of time, energy and money is wasted when people are trying to find a parking lot. These elements could be reduced if the driver is provided vacancy information of a parking lot beforehand. We use pre-trained models such as YOLOv3, YOLOv3-tiny etc. to detect and count the number of cars present in a parking lot. The videos have been captured live from different parking lots. The models are based on Convolutional Neural Network (CNN) with one-stage detection method, which will be explained during the tutorial. Once the model is created, we will deploy it on cloud such as AWS using EC2 instance and create an API endpoint using FastAPI for real time inference.
Outline/Structure of the Demonstration
- Introduction to CNN and different deep learning techniques. - (2-3 min)
- Create a Car Detection model using TinyYolo and pass it through FastAPI - (3-4 min)
- Connect the app to "IPWebcam" to test live using mobile camera - (3-4 min)
- Deploy the app on EC2 instance - (9-12 min)
Understand different deep learning models(especially TinyYOLO)
Create and deploy a model on EC2 instance
Folks who wants to understand deep learning and also deploy it on EC2 instance