EZCap Home

EZCap movement assessments

EZTUG: Timed-up-and-go test

Upload your test. EZTUG tracks your movement and estimates the six phases automatically — no phase-by-phase data entry.

Start seated → stand → walk 3 m → turn → return → sit.

Test setup and safety
  1. Use a stable armchair (seat about 43-46 cm high). Mark a turning point 3 metres from the chair. Wear your usual shoes and use your usual walking aid; record these details below.
  2. Film the entire route with a fixed camera, full body and feet visible. Use a side-on view at the chair if you want trunk-lean estimates.
  3. Start seated with your back against the chair. Record one second at rest. Stand, walk at your usual safe pace to the mark, turn 180 degrees, walk back, turn to the chair and sit down. Stay seated for one second.
  4. Use a fixed side or oblique view showing lateral travel. Include one second seated at each end. Automatic timing starts at movement onset, not the spoken instruction; instruction-timed clinical cut-offs do not apply directly.

Stay safe. Have someone nearby. Stop for pain, dizziness, breathlessness or unsteadiness. Do not attempt this alone if you need assistance. This research tool does not diagnose Parkinson's disease or predict falls.

Stick figure stands from a chair, walks to a marker, turns, walks back and sits. The top-left label changes: stand, walk, turn, walk, sit.
Stand. Walk. Turn. Walk. Sit.

Your assessment

Choose a full TUG video, then select Analyse TUG. Processing stays in your browser.

Recording options

Automatic phase and foot-event estimates are experimental, not clinically validated. Metric lengths, angular velocity and double support are unavailable without suitable calibration or contact observations.

My MP4 will not play or analyse

MP4 is a container, not a video codec. Browser support depends on the codec, device and operating system. If playback fails, re-export the original using H.264 (AVC), 8-bit video, MP4 container (for example with HandBrake: MP4 format, H.264/x264 encoder, not H.265/HEVC). Renaming a file to .mp4 does not convert it. A damaged or incomplete file may need to be exported again from the original recording.

If you use FFmpeg locally, write to a new file to preserve your original:

ffmpeg -i input.mp4 -c:v libx264 -pix_fmt yuv420p -an -movflags +faststart tug-compatible.mp4

This removes audio; EZTUG uses movement onset and does not need the spoken instruction.

AI notice and skeleton replay
Optional timing corrections and calibrated measurements

Automatic analysis does not require these fields. Use them only to correct the detected phases or add verified measurements the video cannot supply.

Correct phase boundaries

Times are absolute video seconds. Every interval includes any pauses within it. Use the same event definition on repeat visits.

Six phases, seven boundaries
Boundary eventVideo time (s)Set from playback

Verified foot contacts (optional)

One row per stance: foot L or R, touchdown time, toe-off time, optional calibrated position along the walking line in metres. Include stances starting before or ending after phase boundaries. List touchdown times in increasing order. Add contacts for both walking phases and both turns.

Example format only: L, 2.100, 2.750, 0.30. Leave position blank if uncalibrated: R, 2.600, 3.250,. Do not paste these illustrative values as your measurements.

Metric definitions and calibration

Step count includes touchdowns in [phase start, phase end). Cadence = count / full phase time × 60, including pauses. Step time uses consecutive opposite-foot touchdowns; stride time uses consecutive same-foot touchdowns. Only pairs wholly within each walking phase are used. Lengths are absolute differences between calibrated foot positions along the straight walking line, not pixel distances. Every eligible pair needs positions; otherwise the length is unavailable.

Double support is overlap between left and right stance intervals, clipped to phase boundaries. We report the mean interval, total overlap and percentage of phase time. Visible feet and accurate toe-offs are essential; an occluded contact must not be guessed.

Use a validated ground-plane calibration (for example a surveyed floor plane and camera homography) or a calibrated external source to obtain foot coordinates and actual walking distances. A single pixels-per-metre scale is not valid for depth-changing perspective. This page accepts calibrated metre values; it does not calibrate camera images automatically. The 3 m course marker is not automatically the distance travelled during each isolated walking phase.

Calibrated heading observations (optional)

To measure angular velocity, supply calibrated body yaw from an external 3D/heading source or reviewed calibrated observations, synchronized to video time. Do not use a 2D shoulder-line angle as body heading. With ordinary video alone, turn time and reviewed step count remain available; angular velocities are unavailable.

Bracket each turn's start and end with observations no more than 0.5 s apart. Unwrap heading across 360 degrees (350, 360, 370, not 350, 0, 10). Mean velocity uses absolute angular travel / duration. Peak is the largest interval-average speed, not an instantaneous instrument peak; denser observations improve its resolution.