Using Yolo V7 Development Of Complete Vids Solution Based On Latest Requirements To Provide Highway Traffic And Incident Real Time Info To The Atms Control Room Using Artificial Intelligence

Abstract

Roads are the most important mode of transport in India. It has a network of over 6,215,797 kilometres of
roads as of 1 December 2021. India has the second-largest road network in the world, followed by the United
States with 6,853,024 kilometres. Physical monitoring of such a large road network is not possible. Even the
police cannot afford to stop the vehicle and fine them as it will lead to massive traffic jams. But this can be
achieved through a smart computerized system which can detect and send e-challans to the violators.
This smart system will analyse the trends and help the government to focus where it needs to increase the size
of roads and create alternative routes to manage heavy traffic. This allows the government to utilize its funds
properly for road development which also benefits the people to get rid of traffic.
Video Incident Detection System (VIDS) is the easiest and smartest way to monitor traffic and check if the
rules are being followed or not. It uses live camera feed to process video frames through the deep neural
networks to identify image features and make decisions accordingly.
To make this system more efficient in these studies we have used the latest YOLOv7 model. Using artificial
intelligence’s deep neural network architecture YOLOv7 model is developed which is used for live object
tracking this model has a higher frame rate and more accuracy than previous YOLO versions. This model has
increased the frame rates from 5 FPS to 160 FPS which is best suitable for monitoring real-time traffic in a
more efficient way. This model outperforms by using GPU-based systems which gives more computational
speed and accuracy.
In these studies, we have proposed models for the vehicle classification system, speed and reverse driving
detection system, helmet detection system, triple seat riding detection system, and number plate recognition
system. This all models will help to make an autonomous traffic monitoring system and send e-challans to
rule violators.