The UE4W dataset was recorded over a period of approximately two weeks by a singe adult male. We found that keeping accurate records and labeling were impractical so it was archived. However, recent advances in representation and unsupervised learning have led us to look at it again. The dataset contains over 250 hours of recordings using an Empatica E4 wristband. In general the longer files are daytime recordings, the shorter ones are nighttime. The tags indicate the beginning and end of an event change (E.g., getting up to go for walk) however the lack of a tag is not a meaningful indicator. A Jupyter notebook is provided which processes each file into a Pandas dataframe and generates a windowed plot of the total acceleration. Additional and updated code may be found on our IMICS lab github. Several reference screenshots from the Empatica Connect Website are also included.

The UE4W dataset was recorded over a period of approximately two weeks by a singe adult male. We found that keeping accurate records and labeling were impractical so it was archived. However, recent advances in representation and unsupervised learning have led us to look at it again. The dataset contains over 250 hours of recordings using an Empatica E4 wristband. In general the longer files are daytime recordings, the shorter ones are nighttime. The tags indicate the beginning and end of an event change (E.g., getting up to go for walk) however the lack of a tag is not a meaningful indicator. A Jupyter notebook is provided which processes each file into a Pandas dataframe and generates a windowed plot of the total acceleration. Additional and updated code may be found on our IMICS lab github. Several reference screenshots from the Empatica Connect Website are also included.

TWristAR is a small three subject dataset recorded using an e4 wristband. Each subject performed six scripted activities: upstairs/downstairs, walk/jog, and sit/stand. Each activity except stairs was performed for one minute a total of three times alternating between the pairs. Subjects 1 & 2 also completed a walking sequence of approximately 10 minutes. The dataset contains motion (accelerometer) data, temperature, electrodermal activity, and heart rate data. The .csv file associated with each datafile contains timing and labeling information and was built using the provided Excel files. Each two activity session was recorded using a downward facing action camera. This video was used to generate the labels and is provided to investigate any data anomalies, especially for the free-form long walk. For privacy reasons only the sub1_stairs video contains audio. The Jupyter notebook processes the acceleration data and performs hold-one-subject out evaluation of a 1D-CNN. Example results from a run performed on a google colab GPU instance (w/o GPU the training time increases to about 90 seconds per pass): Hold-one-subject-out results Train Sub Test Sub Accuracy Training Time (HH:MM:SS) [1,2] [3] 0.757 00:00:12 [2,3] [1] 0.849 00:00:14 [1,3] [2] 0.800 00:00:11 This notebook can also be run in colab here. This video describes the processing https://mediaflo.txstate.edu/Watch/e4_data_processing. We hope you find this a useful dataset with end-to-end code. We have several papers in process and would appreciate your citation of the dataset if you use it in your work.

TWristAR is a small three subject dataset recorded using an e4 wristband. Each subject performed six scripted activities: upstairs/downstairs, walk/jog, and sit/stand. Each activity except stairs was performed for one minute a total of three times alternating between the pairs. Subjects 1 & 2 also completed a walking sequence of approximately 10 minutes. The dataset contains motion (accelerometer) data, temperature, electrodermal activity, and heart rate data. The .csv file associated with each datafile contains timing and labeling information and was built using the provided Excel files. Each two activity session was recorded using a downward facing action camera. This video was used to generate the labels and is provided to investigate any data anomalies, especially for the free-form long walk. For privacy reasons only the sub1_stairs video contains audio. The Jupyter notebook processes the acceleration data and performs hold-one-subject out evaluation of a 1D-CNN. Example results from a run performed on a google colab GPU instance (w/o GPU the training time increases to about 90 seconds per pass): Hold-one-subject-out results Train Sub Test Sub Accuracy Training Time (HH:MM:SS) [1,2] [3] 0.757 00:00:12 [2,3] [1] 0.849 00:00:14 [1,3] [2] 0.800 00:00:11 This notebook can also be run in colab here. This video describes the processing https://mediaflo.txstate.edu/Watch/e4_data_processing. We hope you find this a useful dataset with end-to-end code. We have several papers in process and would appreciate your citation of the dataset if you use it in your work.