Overview / Usage

We developed this project as an initiative for Smart City initiated by Andhra Govt. This pole does what every other pole does. But with some additional features which include anomaly detection and getting vehicle license plate registration if the vehicle violates the traffic rules and automatic registration against the vehicle owner. This is done by a trained model on a raspberry pi which is locally located at the base of each pole like this. Apart from doing computation it receives the data from every pole in that area and predicts the weather(This can have many utilities), This also works as a communication device that has advanced features. In case of emergency, an SOS button is present which sends a notification to drivers and tells traffic ahead and suggests another route to drivers using its huge display which is also controlled by raspberry pi which can be controlled control room.

Methodology / Approach

We have divided the above project into small modules. Like Camera, energy, Emergency, etc. and worked on every module one by one. While solving Licence plate recognition we took dataset from real CCTV cameras available online and trained our model and then tested. After each successful test, we've combined with previously made model and using this approach we made our model which worked on most of the cases. We used a CCTV which has both wired and wireless connection so that in anyone's failure we will have the other case. In the case of wireless connection, we used a raspberry to send video-streams using AWS cloud storage and processing the video data. After processing Video data, it will be sent back to raspberry pi after acknowledging it will decide whether to send data to the control room to take over depending on the case. While all this transmission, a simple small compact sensor will be placed at the top which will send data 24/7 to the raspberry pi which will send to the main control room or central pi to process all this data and send it to AWS to compute and predict the weather. For the convenience of the users, we developed a front-end application to control and see the results on screen so everything can excessed with ease. The server was hosted on AWS.

Technologies Used

Software and Frameworks Used:- Python3, SK-video, Scikit-image, TensorFlow 1.7.0, imgaug, Yolo3, AWS Cloud, CSS, HTML, PHP

Hardware Used:- Raspberry-Pi 3b+ Model, Security Camera, 54Watt battery, Screen, Solar Panel, DHT22 (Humidity, and Temperature sensor), etc

Repository

https://github.com/ksajan/Smart-Pole