Pet motion intelligence API

Inspectable motion data from supported pet footage

Submit supported pet-movement footage to receive pose overlays, frame-level keypoints, confidence values and structured motion measurements. Available outputs depend on the footage, view, visibility and tracking confidence.

Request API access Read the API reference

Pose, keypoint and quality fieldsDocumented, reviewable schema

Structured outputs

Reviewable outputs, from pose to measurements

Pose-overlay video

A reviewable copy of the submitted clip with pose points and skeleton lines where tracking is available.

Frame-level JSON

Coordinates and confidence values for keypoints available in each processed frame.

Motion measurements

Structured stance, stride, joint-angle, velocity, symmetry and posture-derived fields when available.

Stride timeline

Candidate stride-cycle timing and length, with configurable flags for review.

Structured summary

A non-diagnostic summary for an app or professional review workflow.

Explore the schema

Review field names, units, availability notes and proxy limitations before integrating.

Open the field reference

Example response

Inspectable fields, not a single verdict

Metrics use the documented family.field namespace. Body-length-normalised values may support review of the same animal under similar capture conditions; they do not make breeds or recording setups directly comparable.

Example values illustrate the response shape only. Actual fields depend on the submitted footage and tracking confidence.

metrics_v1.json

{
  "scale": {
    "body_length_px": 412.0,
    "motion_axis_stability": 0.94,
    "shoulder_height_bl": 0.61
  },
  "quality": {
    "mean_keypoint_confidence": 0.88,
    "usable_frame_ratio": 0.91
  },
  "stance": {
    "front_left_stance_fraction": 0.62,
    "front_right_stance_fraction": 0.55,
    "front_stance_fraction_asymmetry": 0.12
  },
  "stride": {
    "cadence_steps_per_min": 108.4,
    "mean_stride_length_bl": 1.18,
    "front_stride_length_asymmetry": 0.09
  },
  "head_nod": {
    "head_nod_asymmetry_bl": 0.041
  },
  "weight_bearing": {
    "front_left_weight_index": 0.54,
    "front_right_weight_index": 0.46,
    "front_weight_index_asymmetry": 0.08
  }
}

13 families · 90+ fields

Measurement families in the schema

The schema groups inspectable motion observations by family. Field availability varies with footage, visibility and tracking confidence.

measurement families

90+

motion fields

limb positions represented

scale

Body-length pixel scale, motion-axis angle, capture stability and normalised dimensions.

quality

Keypoint confidence, low-confidence ratio, completeness and usable-frame ratio.

stance

Per-limb stance and swing fractions and timing, plus left/right asymmetry.

stride

Per-limb stride timing, length and consistency, cadence and stride asymmetry.

stride_timeline

Candidate stride cycles with timing, length and configurable review flags.

joint_angles

Per-limb joint-angle mean, minimum, maximum, range of motion and asymmetry.

head_nod

Head-motion asymmetry measurements for professional review, not pain localisation.

hip_hike

Hip-motion asymmetry measurements for professional review, not pain localisation.

velocity

Mean and maximum velocity, variability and steady-state fraction.

back_sway

Lateral sway and vertical oscillation of the back line.

step

Front and back step separation and step width.

symmetry

Front and back paw-lift range asymmetry.

weight_bearing

Per-limb weight-bearing proxy index and front/back asymmetry; this is not a direct force measurement.

Review workflows

Build around observations, not diagnoses

The same pose and measurement primitives can support reviewable products without turning video-derived signals into clinical conclusions.

Professional movement review

Present overlays and measurements for a qualified professional to interpret with history, examination and other relevant evidence.

Within-pet trend review

Compare repeated clips from the same animal only when capture conditions are kept as consistent as practical.

Research workflows

Organise pose and motion observations from supported footage for research datasets, with study-specific quality review.

Product integrations

Connect frame data and documented measurements to developer-owned dashboards, queues and review experiences.

Technical transparency

What the integration exposes

Reviewable pose output

Overlay video and frame-level data make tracking results inspectable rather than reducing them to one score.

Quality alongside measurements

Confidence, completeness and usable-frame fields help indicate when human review may be needed.

Documented measurement layer

The reference exposes metric names, units and proxy limitations without presenting them as medical conclusions.

Structured integration surface

JSON outputs and stride timelines can connect to developer-owned interfaces and review workflows.

Footage and interpretation limits

Output depends on full-body visibility, camera angle, lighting, capture stability, movement speed, occlusion and keypoint confidence. Some fields may be empty or unreliable when footage is unsupported or quality is limited.

These video-derived observations are not pain localisation, a lameness diagnosis, proof of treatment response or an independent clinical conclusion. Qualified professionals should interpret them with history, examination and other relevant evidence.

Plan a responsible integration

Request access to discuss supported footage, schema details, an evaluation setup and the review safeguards your integration needs.