INGENIEUR
Figure 3 : Network connection between IP cameras and AI machine
from roadside IP cameras . The second module is the AI-image processing engine , which comprises vehicle detection , classification , tracking , counting , and anomaly detection algorithms . This module processes the video inputs in real-time and generates data outputs . The third module is the data handler , which manages the output from the AI-image engine either by storing it in the database or sending alerts when necessary . Finally , the fourth module is the dashboard , which displays analytical data in various visual formats , such as bar graphs , pie charts , tables , maps , and time-series line plots .
The AI-image processing engine is the core of the VBITMS system , as it enables the real-time analysis of traffic conditions using advanced algorithms . The vehicle detection algorithm uses computer vision techniques to detect vehicles in the video feed , while the classification algorithm distinguishes between different types of vehicles , such as cars , motorcycles , vans , and trucks . The tracking algorithm tracks the movement of each vehicle over time , while the counting algorithm records the number of vehicles passing through a specific location . The anomaly detection algorithm monitors the video feed for any abnormal behaviour , such as illegal parking or accidents .
Figure 5 displays examples of real-time vehicle detection and classification using AI algorithms , which can assign unique identities to different vehicles . The system also incorporates a region filter ( represented in red ) that allows for targeted measurements in specific areas . As seen in Figure 5 ( a-d ), the AI algorithm-based detection successfully identifies various types of vehicles , including buses , trucks , and motorcycles .
66 VOL 93 JANUARY-MARCH 2023