Vehicle detection and classification using acoustic and seismic data

In this work, co-operative detection and classification algorithms from measured sound and vibration produced by ground vehicles are presented. Vehicle classification utilizing acoustic and seismic data enables the identification of vehicle types through analysis of sound and vibration characteristics. Machine learning models employed include CNN, LSTM, SVM, and federated learning. The SVM, KNN, and CNN models performed exceptionally well on the test set, with accuracies ranging from 98.78% to 99.76%. The KNN model achieved the highest accuracy of 99.76%, while the CNN model demonstrated a strong performance with a slightly lower accuracy of 99.51%.