Skip to main content
All CollectionsUsing Uplift Capture
Interacting With Kinematic Data
Interacting With Kinematic Data

How to interact with the kinematic data generated by Uplift Capture.

M
Written by Matthew Kowalski
Updated over a year ago

There are three main ways for users to interact with our kinematic data: our cloud visualizer, the reports we generate for certain activities and movements, and our CSV exports.

Cloud Visualizer

Any recorded session’s data can be accessed online with the cloud visualizer. Our cloud visualizer is currently under development, and only the 3D avatar, as well as the Kinematic Sequence and X-Factor graphs, use our updated models. For more on interacting with the cloud visualizer, click here.

Activity Reports

For select activities, data computed from our pipelines is aggregated into a single summary report. Please see the following for details and examples of our summary reports:

We also currently generate reports for the following Movement Assessment movements:

  • Bodyweight Squat

  • Single Leg Squat

  • Overhead Squat

  • Countermovement Jump

  • Single Leg Jump

  • Paused Squat Jump

  • Lunge

Find more details on our movement assessment reports here.

CSV Exports

In addition to our joint center and angle data, our CSV exports also provide additional information about a session, such as:

  • Athlete name

  • Athlete id

  • Session id

  • Org id

  • Capture time (the time, in seconds, since the Unix epoch, which is January 1, 1970, 00:00:00 UTC)

  • Frames per second (fps)

  • Activity tag

  • Movement tag

  • Handedness

  • Footedness

  • QA tag (defaults to 1 except for certain activities and movements where the Uplift Team has built a dedicated QA process- a good “hint” as to whether our team has created a dedicated QA process is if there are additional events and metrics included in the export)

  • Time (in seconds)

  • Frame (number)

  • Any additional events and metrics based on the activity and movement selected (e.g., baseball pitching will also export a string containing the kinematic sequence ordering under the column kinematic_sequence_order)

These data exports provide the user with the flexibility to ingest the data to conduct their own bespoke analyses that we may not yet support.

“Approved by Biomech QA”

We often receive questions about what our “approved by biomech QA” column means in our CSV exports. Depending on the task, we check different aspects of the data to ensure that data quality is adequate for building our automated reports. Therefore, it is possible that “not approved” data may actually meet the requirements for your specific research questions and vice versa. Please note that any movements that do not pass QA will not be included in generated PDF reports.

The main “quality” checks we currently have in production are as follows:

Baseball Pitching

  • Whether the pelvis, trunk, and arm velocities in the kinematic sequence are within a certain range (too low and we assume no pitch, too high and we assume there's some other noise in the data)

  • Whether the angles themselves are "realistic"

  • Whether we've detected all events

  • Whether the events we've detected are within a reasonable window

Baseball Hitting

  • Whether the pelvis, trunk, and arm velocities in the kinematic sequence are within a certain range (too low and we assume no swing, too high and we assume there's some other noise in the data)

  • Whether the angles themselves are "realistic"

  • Whether we've detected all events

  • Whether the events we've detected are within a reasonable window

Countermovement Jump, Single-Leg Jump, and Squat Jump

  • Whether we have detected all events

  • Whether the events occurred in order

  • Whether the jump height is realistic

  • Whether the hip-knee dominance value is within range of 0-1

  • Whether the events we've detected are within a reasonable window

Bodyweight Squat, Single-Leg Squat, Overhead Squat, and Front Lunge

  • Whether we have detected all events

  • Whether the events occurred in order

  • Whether the hip-knee dominance value is within range of 0-1

  • Whether the events we've detected are within a reasonable window

Note that we may update these QA algorithms over time depending on user requests or updates to our models.

Data Smoothing

Please note that all data exports have been smoothed with a zero-lag, (effective) fourth-order low-pass Butterworth filter. The cutoff frequencies for this filter were based on residual analyses outlined by David Winter (1990).

Did this answer your question?