Dataset of human gait in a domestic setup collected with an LFMCW MIMO radar
doi: 10.4121/7cfe7f4f-e8ef-4a06-8359-ca5831e22810
Radar technology is a strong candidate to analyze human gait. This dataset contains gait captures of 60 healthy individuals in a laboratory emulating a domestic scenario. The data is captured with the 24 GHz LFMCW MIMO radar Inras RadarBook 2 and with the Microsoft Azure Kinect camera. All the measurements are recorded with the approval of the ethics committee of TU Delft, under the project DARE-NERD. For further information regarding the radar and the Kinect configuration, and the experiment conditions, please refer to an article currently submitted for review.
The dataset contains a ".zip" file for each participant with:
- 3 ".mat" files, corresponding to each trial performed by the subject. Each file contains the radar data in the variable "tima_data". The dimensions are (fast time) x (slow time) x (channel index). Moreover, the radar configuration is contained in the variable "Ana".
- 3 ".csv" files, corresponding to each trial performed by the subject. Each file contains the 32 joint locations extracted with the Microsoft Azure Kinect camera for each time frame.
- A ".csv" file with the time shifts that need to be applied to the Kinect data to synchronize it with the radar data.
Moreover, the dataset provides
- An ".xlsx" file containg the identifier, gender, age, height, weight and shoe size of each participant.
- A matlab script to extract the radar data from the ".mat" file.
- A matlab script to extract the joint data from the ".csv" file.
The authors would like to thank the participants in the experiment for their availability.
- 2024-10-25 first online
- 2024-10-30 published, posted
Microwave Sensing, Signals and Systems (MS3) Group, Delft University of Technology