FIGURE 4 - available via license: Creative Commons Attribution 4.0 International
Content may be subject to copyright.
![Original and CycleGAN-generated Data. Shown is the test set data of Participant 3 (1) and Participant 5 (2). Panels 1.1 and 2.1 show the original walking data and the genRunning and genHandwriting data generated from them. Accordingly, panels 1.2 and 2.2 refer to the original running data, and panels 1.3 and 2.3 to the original handwriting data each with the data generated from them. The mean values and standard deviation over the respective curves are shown in each case. On the x-axis, the respective courses over the 256 time points are shown. The y-axis shows the vertical force scaled to the interval [0, 1]. For the walking and running data, the values 1 to 128 represent the ground contact of the left foot and the values 129 to 256 of the right foot. genRunning, generated running data; genWalking, generated walking data; and genWriting, generated handwriting data.](publication/372982537/figure/fig2/AS:11431281180182595@1691507425253/Original-and-CycleGAN-generated-Data-Shown-is-the-test-set-data-of-Participant-3-1-and.png)
Original and CycleGAN-generated Data. Shown is the test set data of Participant 3 (1) and Participant 5 (2). Panels 1.1 and 2.1 show the original walking data and the genRunning and genHandwriting data generated from them. Accordingly, panels 1.2 and 2.2 refer to the original running data, and panels 1.3 and 2.3 to the original handwriting data each with the data generated from them. The mean values and standard deviation over the respective curves are shown in each case. On the x-axis, the respective courses over the 256 time points are shown. The y-axis shows the vertical force scaled to the interval [0, 1]. For the walking and running data, the values 1 to 128 represent the ground contact of the left foot and the values 129 to 256 of the right foot. genRunning, generated running data; genWalking, generated walking data; and genWriting, generated handwriting data.
Source publication
p>In recent years, the analysis of movement patterns has increasingly focused on the individuality of movements. After long speculations about weak individuality, strong individuality is now accepted, and the first situation–dependent fine structures within it are already identified. Methodologically, however, only signals of the same movements hav...
Contexts in source publication
Context 1
... presented in Table 3, it is possible to distinguish the participants from each other in both the walking and running data, and the handwriting data with more than 98.0% classification F 1 -score each. Figure 4 shows an example of the data of the participants p3 (Figure 4.1) and p5 (Figure 4.2). Looking at the shape of the curves, it is noticeable that the generated running (genRunning), generated An SVM with five-fold cross-validation was applied. ...
Context 2
... presented in Table 3, it is possible to distinguish the participants from each other in both the walking and running data, and the handwriting data with more than 98.0% classification F 1 -score each. Figure 4 shows an example of the data of the participants p3 (Figure 4.1) and p5 (Figure 4.2). ...
Context 3
... presented in Table 3, it is possible to distinguish the participants from each other in both the walking and running data, and the handwriting data with more than 98.0% classification F 1 -score each. Figure 4 shows an example of the data of the participants p3 (Figure 4.1) and p5 (Figure 4.2). Looking at the shape of the curves, it is noticeable that the generated running (genRunning), generated An SVM with five-fold cross-validation was applied. ...
Context 4
... addition, the examples of genRunning data from walking data show a small wave at the end of the left and right ground contact, which does not occur in the original running data. In the genWalking and genRunning data, which are based on the handwritten data, it can also be seen that there are sections within the curves that vary noticeably in variance (e.g., Figure 4.1.3: the right ground contact in each case for genWalking and genRunning). ...
Context 5
... more detailed look at the genWriting data shows that they deviate somewhat more from the original data than genRunning data and genWalking data. In addition, for example, there are also fewer smooth curve components (Figure 4.1.2). Furthermore, it is noticeable that the first value of the generated data tends to be too small (i.e., by 0.0) and does not match the original data (i.e., by 0.2). ...
Context 6
... example, in both p3 and p5, the somewhat stronger impact peak of the right ground contact during running is also transferred accordingly in the respective genRunning data. In addition, in the genWalking data at p3 (Figure 4.1.2), the left ground contact shows a higher loading peak than the terminal stance peak, which is exactly the opposite in the right ground contact but corresponds to the original data in both cases. ...
Context 7
... left ground contact shows a higher loading peak than the terminal stance peak, which is exactly the opposite in the right ground contact but corresponds to the original data in both cases. Also, in handwriting, for example, it can be seen that the pattern of p5 (Figure 4.2.3) is somewhat wavier than that of p3 (Figure 4.1.3), which is also reflected in the corresponding genWriting data. ...
Context 8
... generation of the genRunning (92.5% F 1 -score) and genWalking (98.9% F 1 -score) data from the original walking and running data worked particularly well. This impression can be confirmed by looking at the figures (Figure 4) of the generated data. The generated genWalking and genRunning data from the original walking and running data are not only very similar in their general shape to the original data but also reflect the respective individual characteristics such as impact peak or time course features in their curves. ...
Context 9
... generated genWalking and genRunning data from the original walking and running data are not only very similar in their general shape to the original data but also reflect the respective individual characteristics such as impact peak or time course features in their curves. However, the figures (Figure 4) also show differences between generated and original data. For example, the variance of the generated data was shown to be significantly lower than that of the original data. ...
Context 10
... the results are worse than those of the genRunning and genWriting data generated from the walking and running data. The figures (Figure 4) provide a first explanation in this respect. At first glance, the genRunning and genWalking data from the handwriting data and the genWriting data from the running and walking data correspond quite closely to the original data, including the adoption of the curve-specific individual characteristics. ...
