A model is still running

A model is still running or downloading. Leaving will stop the current operation. Stay on this page to let it finish.

Skip to content
LocalAI.toolsBeta

Join the waitlist

Get an email when the new apps and model runs launch.

No account needed.

Back to models

MediaPipe Hand Landmarker

Locate hands and 21 landmarks on each detected hand in a still image. Returns image and world-coordinate landmarks plus left/right-hand predictions; it does not recognize sign language.

Browser inferencePowered by WebAI · Private · Unlimited runs
About this model

Locate hands and 21 landmarks on each detected hand in a still image. Returns image and world-coordinate landmarks plus left/right-hand predictions; it does not recognize sign language.

Created by
Google
License
Apache-2.0
Model released
Unknown
Source repository created
Jun 13, 2019
Model weights · FP16
7.8 MB

Best for

  • Gesture interfaces and hand-overlay prototypes.
  • Extracting hand geometry for non-consequential interaction or analysis.

Keep in mind

  • Hands that are small, cropped, occluded, blurred, poorly lit, or in unusual poses can be missed or receive inaccurate landmarks.
  • Left/right handedness can be unreliable for mirrored or ambiguous images.
  • World landmarks are model estimates, not calibrated physical measurements.
  • Do not use landmarks or handedness for identity, medical, safety, or other consequential decisions.

Processed on your deviceInference runs in this browser. Input/output sharing requires consent for each run. Basic run logs include device information and approximate connection location. Model and runtime files download from external hosts. Saved downloads may be cleared by your browser.

Sources & technical details
Original model
google-ai-edge/mediapipe
Release verification
Unknown
Release evidence
Exact original checkpoint public release date has not been verified. Repository creation, training, paper, artifact conversion and update dates are not release evidence.
Source updated
Apr 26, 2023
Browser model files
mediapipe-models/hand_landmarker/hand_landmarker
Files updated
Apr 26, 2023
Artifact revision
hand-landmarker-float16-generation-1682480005356399

Input

IMAGE

Sample result

MediaPipe Hand Landmarker

Settings

Maximum hands returned. Lower values limit output to fewer hands. Higher values allow more hands to be processed and can increase work; they do not create detections when no hands are visible.
Adds a self-contained SVG presentation image alongside the unchanged native model fields. When enabled, draws model predictions or presents generated media and text. Adds CPU postprocessing time and image bytes. It does not change model predictions.
Advanced settings 2
Sets the confidence required to accept a hand detection. Lower values accept more uncertain hands and possible false positives. Higher values can reject partially visible or difficult hands.
Sets the confidence required for hand presence in the landmark stage. Lower values accept less certain landmark results. Higher values reject uncertain hands and may reduce the number of returned landmark sets.

Output

Your result will appear here

Add input or try the sample, then run the model.

Benchmark evidence

Saved worker evidence is separate from benchmarks run on your device. It does not guarantee speed or quality on every browser.

MediaPipe Hand Landmarker

1 configurations tested
Passed1
Failed0
TestedSep 9, 2026
RuntimeMediaPipe Tasks Vision 1.0.1

Quality criteria

No explicit quality gate recorded. Review per-case evidence; full certification is pending.

PrecisionDeviceResult
FP16CPU · WASMPassedVery fast
Test setup & sample details
Quality gate
No explicit quality gate recorded. Review the per-case quality evidence; full certification is pending.
Tested
2026-09-09T06:08:00.000Z
Environment
Codex in-app Chromium 151 on macOS; MediaPipe Tasks Vision 1.0.1 WASM CPU
Runtime
MediaPipe Tasks Vision 1.0.1
Artifact revision
hand-landmarker-float16-generation-1682480005356399
Fixture
{
  "file": "mediapipe-hand-test.jpg",
  "source": "https://raw.githubusercontent.com/google-ai-edge/mediapipe-samples/7f3cf17410db91f8e084a46f7517df15bce3479a/examples/hand_landmarker/android/app/src/androidTest/assets/test_image.jpg",
  "url": "https://f005.backblazeb2.com/file/hubters-webai-public/assets/benchmarks/mediapipe-hand-test.jpg",
  "size": 33051,
  "sha1": "d44bc4fbfbc8587039ab32df880162dca7fe842e",
  "width": 640,
  "height": 427,
  "license": "Apache-2.0",
  "licenseUrl": "https://github.com/google-ai-edge/mediapipe/blob/master/LICENSE",
  "sourceRevision": "google-ai-edge/mediapipe-samples@7f3cf17410db91f8e084a46f7517df15bce3479a"
}
Rerun when
model artifact generation change; MediaPipe Tasks Vision version change; fixture or quality-gate change; browser/backend behavior change
Per-case quality & failure details
[
  {
    "precision": "fp16",
    "device": "wasm",
    "status": "pass",
    "quality": {
      "grade": "pass",
      "expectedHandCount": 1,
      "actualHandCount": 1,
      "imageLandmarkCount": 21,
      "worldLandmarkCount": 21,
      "handedness": "Right",
      "handednessScore": 0.703342080116272
    },
    "runtime": {
      "loadMs": 2261.4,
      "inferenceMs": 82.6,
      "totalMs": 2344,
      "grade": "excellent"
    }
  }
]

Something not quite right?

Tell us what happened with MediaPipe Hand Landmarker. Loading, results, or the page itself—we’re here for all of it.

Attachments Optional

Up to 3 files · 50 MB each

Browser, OS, CPU/GPU, memory, screen, storage, and network details are included automatically. Model inputs, results, and browsing history are not collected.