Transfer Learning based Emergency Vehicle Detection on Traffic Signal through IP Camera

The traffic congestion leads to high impact for emergency vehicles to pass by the particular road. The congestion can be deleted by many ways where in the video analytics-based detection and give more accuracy and able to count the type of vehicle in the particular traffic signals. The early detection of emergency vehicle in the particular signal can be avoid delay in transition. To overcome this problem, the proposed model that detects the emergency vehicle in the signal and pass the information to the signal regulating authorities to green corridor can be generated have on the information extracted from the video signals. The model constructed with the 5 layers with the base of resnet50 architecture. The hyperparameter are tuned to obtain the maximum accuracy and minimum loss. The parameter such as precision and recall where evaluated to compare the existing Resnet50 and modified Resnet50.