The Description of the StrengthSense Dataset

The StrengthSense dataset is designed for detecting daily behaviors and focuses on common activities that involve strength in everyday life. Ten IMU sensors were employed to record IMU9 data. These sensors were strategically placed on the subject's body: one on the chest, one on the waist, two on each upper and lower arm, and two on each thigh and calf. Data was collected from 29 participants aged 18 to 60, covering 13 typical daily activities. These activities include:

1) walking, walking on an incline, walking on a decline
2) sit-to-stand from chairs with/out handles
3) walking while carrying one/two shopping bags (2 to 5kg/bag)
4) mopping floor
5) lying on an exercise mattress
6) sitting on sofa
7) lying on sofa
8) greeting (waving hands)
9) opening a bottle of water, pouring water to a glass and drinking from the glass
10) climbing up/down stairs
11) push up
12) sit up
13) squat

The data files are organized by subject (e.g., subject 1, 2, 3, ...). Each subject's folder contains two subfolders, laptop1 and laptop2, which separately store data for the upper body (including the chest and arms) and the lower body (including the waist and legs). Five IMU sensors are used in the recordings in the laptop1 and laptop2 folders respectively.

Within each laptop folder, there is a subfolder: IMU9. The IMU9 folder contains data recorded by the IMU sensors in CSV format.

The file naming convention follows the format: s##_a##_t##_@##.suffix, where:

• s## refers to the subject number,
• a## denotes the activity category,
• t## indicates the trial number,
• @ specifies either u or l for upper body or lower body data,
• ## represents the respective Arabic numeral,
• and suffix refers to the file extension (.CSV).

Access to and use of this dataset is granted for non-commercial research purposes only. Commercial use of the dataset is strictly prohibited. Access to and use of this dataset is permitted only for users who are physically located within the European Union member states.

*Note: Due to Bluetooth disconnection during data collection, only about the last 10 seconds of Subject 20’s data are usable. It is therefore recommended to exclude this subject from analysis. In addition, for a very small number of trials from some subjects and activities, Bluetooth connectivity issues also occurred, resulting in a recall of 0. These cases should likewise be treated with caution during use.