Emergency vehicles classification for traffic signal system using optimized transfer DenseNet201 model

<span>As a result of the rapid growth world population, traffic signaling systems for monitoring and controlling roads have turned to be an important issue facing humanity. To effectively overcome this problem, accurate method congestion reduction on should used which has direct relation between population cars’ usage. Various approaches derived from deep transfer learning been investigated in context. This research implemented optimized approach densely connected convolutional neural network (DenseNet201) models multiple classifications (non-emergency cars, ambulance, police, firefighter). Due non-availability public datasets, customized dataset created. paper aims improve performance accuracy vehicle classification using certain preprocessing algorithms input images testing various optimization methods. The proposed model is evaluated k-folds cross-validation 20:80 test training, respectively. metrics are comparison with other techniques based exactness, recall, F1-score. Test outcomes specify that model-based outperforms alternative regarding classifying reaches 98.6%.</span>