Figure 12 | Scientific Reports

Figure 12

From: Reliability and generalization of gait biometrics using 3D inertial sensor data and 3D optical system trajectories

Figure 12

Bi-dimensional UMAP projections using the feature vector pairs generated from accelerations and marker trajectories filtered at different frequencies. Four folds of pairs were generated: training set of same-subject pairs, training set of different-subjects pairs, test set of same-subject pairs, test-set of different-subjects pairs. The band-pass filtering from 6 to 10 Hz lead to the greatest separability for the pairs. The train and test sets consistently overlap, indicating a robust learned representation.

Back to article page