There has been a population increase which has consequently led to traffic congestion in the city of Karachi. Making a smart traffic management system that makes use of video and picture data of the traffic on the roads of Karachi, Pakistan. This works by performing machine learning using an algorithm over the recent frame obtained from the video to estimate the number of vehicles present in a scene. Cameras will be installed on the opposite of the lane, beside the traffic light and will take its real-time video. At the back-end, Raspberry Pi would be connected to handle video processing. Raspberry pi would receive video as input from the camera of each road. Image framing would capture frames from the video at several fixed intervals. By taking our city, Karachi, into consideration we are creating our data set based on images captured from within the city. The proposed project aims to make decisions for the traffic signal timings based on vehicle densities. The project will be deployed on a four-way traffic signal. It will make use of image processing to separate image frames while machine learning algorithms will perform the task of signal controlling and vehicle detection. The reason for using Image Processing and machine learning is because it keeps production costs are low while achieving high speed and accuracy.
For more details on this project, check out my blog on medium: https://tubasid.medium.com/smart-traffic-management-system-939579031bee Get annotated dataset on Kaggle: https://www.kaggle.com/tubasiddiqui/toy-cars-annotated-on-yolo-format