Pose and joint analysis · AI beta
Pose and joint analysis · AI beta — Photo / video keypoints / PNG / CSV / JSON
One photo or browser-decodable MP4 / WebM · 64 MiB / 30 seconds. MoveNet Lightning estimates 17 keypoints for one person. Analysis images have a max side of 640 px; video is sampled at 1 fps, up to 30 frames. Save the first annotated frame PNG and all coordinates CSV / JSON. No medical / posture diagnosis or repetition counting.
- Select video
- Check range and output settings
- Process, play and save
Usage example
- Example input
exercise.webm [10 s]- Expected result
First annotated frame + 1 fps keypoint coordinates
Examples are illustrative. Actual results depend on the source file and settings.
Examples and limitations
Photo / video keypoints / PNG / CSV / JSON
Input and output
- Supported input
image/png · image/jpeg · image/webp · .png · .jpg · .jpeg · .webp · video/mp4 · video/webm · .mp4 · .webm- Saved output
PNG · CSV · JSON
Usage example
- Example input
exercise.webm [10 s]- Expected result
First annotated frame + 1 fps keypoint coordinates
Examples are illustrative. Actual results depend on the source file and settings.
Limits and checks
- One photo or browser-decodable MP4 / WebM · 64 MiB / 30 seconds. MoveNet Lightning estimates 17 keypoints for one person. Analysis images have a max side of 640 px; video is sampled at 1 fps, up to 30 frames. Save the first annotated frame PNG and all coordinates CSV / JSON. No medical / posture diagnosis or repetition counting.
- Models and runtimes download from this Site. Selected files are never uploaded. Review AI predictions before saving.