How close are you to nailing the pose? Guessing gets you nowhere. Pose Match uses TensorFlow.js MoveNet to detect 17 body keypoints through your camera and compares them against the target pose in real time — giving you a live percentage score.
Score above 80% and you're in the zone. Here's how to use it.
Pose Match runs a TensorFlow.js MoveNet model directly in your browser — no data leaves your device. The model detects 17 body keypoints (nose, shoulders, elbows, wrists, hips, knees, ankles) and maps them to our 15-point skeleton system.
It then compares the bone vectors (the angles between your joints) against the target pose using cosine similarity, with heavier weighting on torso and head positioning. The result is a real-time percentage score displayed on screen.
How to Use It
Step 1
Select any pose and tap Pose Match
Choose a pose, then open the Feature Hub (⚡ icon) and tap 🎯 Pose Match. The AI model loads in a few seconds — you'll see a loading indicator the first time.
Step 2
Position yourself in frame
Stand back far enough so your full body is visible. The model needs to see your major joints. A live skeleton overlays your body showing which keypoints are detected.
Step 3
Match the pose and watch your score
Adjust your body to match the overlay. The score updates in real time. Green (80%+) means you're matching well. Capture when you're satisfied — the score is saved with your photo.
Score 80%+ on the Daily Challenge pose to complete it and build a streak!
How Scoring Works
The scoring algorithm compares bone vectors — the direction and angle between connected joints — using cosine similarity. Torso and head connections are weighted 2x more than limb extremities, because core body position matters most for a recognizable pose.
The score is a percentage from 0 to 100. Getting above 80% means your body position is visually matching the target pose. Above 90% is photographer-level precision.
Tips for Higher Scores
Good lighting is essential
The AI model detects your body through your camera. Poor lighting means lower detection confidence, which means lower scores. Natural light or a well-lit room works best.
Full body in frame
Pose Match needs to see your ankles, knees, hips, shoulders, and head. If you're cut off at the knees, some keypoints can't be detected and your score will suffer.
Focus on torso first, then limbs
Since torso and head are weighted more heavily, get your core body position right first. Then fine-tune your arm and leg positions to push the score higher.
FAQ
Does Pose Match send my camera data anywhere?
No. The TensorFlow.js model runs entirely in your browser. No images or body data leave your device.
What score should I aim for?
80% or above means you're matching the pose well. 90%+ is professional-level precision. Even 70% usually looks good in a photo.
Why is my score low even though I look right?
Make sure your full body is visible and the lighting is good. The model needs clear visibility of your joints to score accurately.
Does it work on all phones?
Yes — it works on any modern browser with camera access. Newer phones will run the model faster.
Try Pose Match Now
Open PoseOverlay in your browser — no download, no signup.