Bike type identification using smartphone sensors

Deep Learning-Based Vibration Signal Personnel Positioning System

“…Considering that existing works [15], [34], [35] suggest that deep learning is more suitable to analyze time-series of sensor data, our approach is based on CNNs as well. The works [34] compared CNN, LSTM, perceptron, and random forest and conclude that CNN provides the best results. The existing works [35] use the same sensors on a pattern recognition task and compare a deep learning approach with random forest classifiers in the experiment.…”