Hi, It is not easy to see an IMU dataset! Thanks for sharing this!

For my use case, I plan to transfer the data to SMPL format so that I can use it with other data in SMPL, like AMASS.
I mapped the Xsens to SMPL by using the joints on the body directly. (Remove L5 joint)
When I used the quaternion to visualize the mesh, the mesh's distorted.
After processing the data by preprocessing.py and quaternion.py before visualization, the distortion disappeared.
However, the motion of the mesh still looks weird. (Limbs are uncoordinated.) Can anyone give me some help or suggestion?

Thanks in advance!

Hi Dr.Geissinger,

I am a graduate student in Computer Science at McMaster University. I found that Viginia Tech Natural Motion dataset is replicable in the Human Pose Estimation field when reading your paper Motion Inference Using Sparse Inertial Sensors, Self-Supervised Learning, and a New Dataset of Unscripted Human Motion. However, using this link (https://data.lib.vt.edu/articles/dataset/Virginia_Tech_Natural_Motion_Dataset/14114054) will only lead to the h5_P1_Day1 data but no other subjects, and Globus always times out when Retrieving Directory Contents. Could you please introduce other ways to download the complete datasets or check on these two approaches? Thank you so much for the work. Looking forward to seeing the whole dataset!

Xijian