Study of Existing Algorithms & Implementation of Vehicle Detection for Smart Warning System for Drivers

It is essential to detect accident for efficient transportation system. Many accidents occur at sharp turns of Ghats or on highways. In case of emergency situations some drivers park their vehicle alongside highways or Ghats and it provokes accidents. This paper studies deep learning techniques for vehicle detection. Smart warning system for drivers is the focused application implemented in this paper. In previous papers only vehicle detection to prevent accidents is done for vehicles in motion. Whereas in this paper, both parked vehicles and in motion vehicles are detected in real time. System detects cars, motorbikes, trucks and buses and notifies the driver with alarms, message notifications or seat vibrations. To implement this system, dataset from Kaggle named Vehicle Detection image set is used. 7036 samples are been used to train the model and 1760 samples to test the model. Model's average accuracy of detection is 82.5%.