QValiData Experiment Data

Data set for validation experiments performed using the newly-developed QValiData video validation and simulation software. This archive contains:

1. Data sets for all five video validation trials
2. R scripts for generating figures used in the manuscript
3. Final results, organized in CSV form

Update 2 July 2021: Add new post-processing files for new figure style

Directory Structure:

• experiment-data
  • test*
    • Contains QValiData configuration file, activity logger data, and video recordings for all tests involved in the experiment
• post-process
  • data: Contains data files describing activity detector performance. Each file name is given in the following format: [bit|val]_halfsecond_[nowakeup|wakeup|]_[trial_number]_[activeevents|videoevents]
    • [bit|val]
      • bit: Data set uses "Bit Tag" activity detection parameters, developed prior to the implementation of the new methodology.
      • val: Data set uses parameters found experimentally using simulation-based validation.
    • halfsecond: The "ground-truth" data set used (list of events classified as "active" by video) was trimmed to only include events lasting half a second or longer (15 or more frames of video)
    • [nowakeup|wakeup]
      • nowakeup: ADXL362 low-power activity wakeup mode was not used for this trial.
      • wakeup: wakeup mode was used for this trial.
    • [trial_number]: A number in the range 1-5 indicating which data set the results came from
    • [activeevents|videoevents]
      • activeevents: The file contains a list of all events detected by the simulated activity detector. Each row contains a number indicating the length of each event (in seconds) and a TRUE/FALSE value that is TRUE only if the event contains at least one sample that was also marked as "active" by video analysis.
      • videoevents: The file contains a list of all events detected by video analysis. Each row contains a number indicating the length of each event (in seconds) and another number indicating the percentage of each trial that was also marked as "active" by the simulated activity detector.
  • correctevents.R: Plots histograms of events detected via video analysis that were also detected by simulated activity detection using each set of activity detection parameters.
  • falsepositives.R: Plots histograms of events detected via simulated activity detection, but were not detected by video analysis, which reveals the occurrence of false-positive classifications.
  • correctevents_new.R: Same as previous version, but with reworked figures.
  • falsepositives_new.R: Same as previous version, but with reworked figures.
  • post-process.Rproj: R project file for both R scripts.
• QValiData-Parameters.csv: Contains all activity detection parameters used for the five trials.
• QValiData-Results.csv: Contains simulation performance metrics for all five trials.

This material is based upon work supported by the National Science Foundation under Grant No. 1644717. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Science Foundation.