Fencing Footwork Dataset

The dataset is designed for analysis of action dynamics. It contains 6 fencing footwork actions: Four lunge actions have similar average trajectories, but different dynamics of the movement, therefore classification of these actions must be based on dynamics.

The dataset includes samples from 10 fencers and was recorded due to the courtesy of Aramis Fencing School aramis.pl

Samples were recorded with the x-IMU 9 axis inertial sensor as well as the Kinect sensor.

Provided data includes accelerometer, gyroscope, magnetometer and rotation data acquired with the x-IMU, as well as depth and skeleton data acquired with the Kinect. RGB data is not included.

Format of the data is as follows:

  • Kinect skeleton data: *_Body.mat.
    MATLAB format, array n x 161, where n is the number of frames. For reach frame there is a timestamp and 8 values for each of 20 joints tracked by the Kinect. Values for a single joint include: tracking status, x,y,z positions in camera coordinate system, x,y positions in color coordinate system, x,y positions in depth coordinate system.

  • Kinect depth data: *_Depth.avi.
    Saved as RGB video, where depth data (16bit) is saved in R (less significant 8 bits) and G (more significant 8 bits) channels. To prevent data loss the video is not compressed. For more information on Kinect depth stream see: https://msdn.microsoft.com/en-us/library/jj131028.aspx

  • IMU data: *_IMU.mat.
    MATLAB format, array n x 10, where n is the number of frames. For each frame there is a timestamp and 9 values: acc x,y,z, magn x,y,z, gyro x,y,z.

  • IMU rotation data: *_Rot.mat.
    MATLAB format, array n x 17, where n is the number of frames. For each frame there is a timestamp and rotations computed by the IMU's internal software, given as quaternions (4 values), euler angles (3 values) and rotation matrix (9 values).

    The dataset can be downloaded here: fencing footwork dataset . Total size of zipped files is approx. 1GB. After unzipping total size of the dataset is over 30GB.

    For more information please contact Filip Malawski: fmal[at]agh.edu.pl