VIDIMU-TOOLS is a code repository related to the public dataset "VIDIMU. Multimodal video and IMU kinematic dataset on daily life activities using affordable devices."
- The VIDIMU dataset can be freely accessed in Zenodo repository (doi: )
- The scientific paper can be freely accessed in Scientific Data (Nature) (doi:)
The processing of raw data in the VIDIMU dataset was done using the scripts included in this repository, in combination with free tools NVIDIA Maxine SDK BodyTrack (v0.8) and OpenSim (v4.4).
- Download the VIDIMU dataset to a folder in your local machine.
- In the desired Jupyter notebook included in VIDIMU-TOOLS, set the
fulldatasetpathvariable to the path of the previous folder.
The code is organized in the following folder hierarchy:
- imus folder includes the following Jupyter notebooks:
PlotImusRawQuats.ipynb: generates .svg plots of the raw quaternion data acquired with custom IMU sensors and stored as .raw files in the VIDIMU dataset.PlotImusIkJointAngles.ipynb: generates .svg plots of the joint angles estimated through Inverse Kinematics in OpenSim and stored as .mot files in the VIDIMU dataset.
- video folder includes the following Jupyter notebooks:
ScriptsToBodyTrack.ipynb: generates a list of video files stored as .mp4 in the VIDIMU dataset that can be used to process them with NVIDIA Maxine SDK BodyTrack.ConvertBodytrackToCSV.ipynb: converts the plain text output of NVIDIA Maxine SDK BodyTrack stored as .out files in the VIDIMU dataset, into comma separated values .csv files.PlotVideoEstimatedJointAngles.ipynb: generate plots the joint angles estimated from 3D joint positions inferred by NVIDIA Maxine SDK BodyTrack by reading .csv files.RecodeMP4toSmallsizefiles.ipynb: recodes original acquired and NVIDIA Maxine SDK BodyTrack generated .mp4 video files to significantly reduce their size and stores them in a different folder.
- synchronize folder includes the following Jupyter notebooks:
EstimateFileSynchronization.ipynb: computes ideal synchronization of IMU and video data records by estimating RMSE of shifted joint angles signals (.mot and .csv files), and writes this info to a fileinfoToSync.csv.ModifyFilesToSync.ipynb: modify VIDIMU dataset files for estimated ideal synchronization according toinfoToSync.csv.
- utils folder includes auxiliary Python functions employed in the Jupyter notebooks commented above.
When using this code, please include a reference to this GitHub repository and the associated scientific paper https://arxiv.org/abs/2303.16150:
Martínez-Zarzuela, M., González-Alonso, J., Antón-Rodríguez, M., Díaz-Pernas, F. J., Müller, H., & Simón-Martínez, C. (2023). VIDIMU. Multimodal video and IMU kinematic dataset on daily life activities using affordable devices. arXiv preprint arXiv:2303.16150.