AI-powered pose estimation tool using MediaPipe. Upload an image to automatically detect 33 body landmarks, measure joint angles, and analyze posture.
이미지를 드래그하거나 클릭하여 업로드
JPG, PNG, WebP 지원Upload an image containing a person you want to analyze. You can drag and drop or click to select a file.
Click the Detect Pose button and AI will automatically detect 33 body landmarks.
A skeleton will be drawn on the image, and you can check joint angles and posture analysis results.
Download the annotated image with landmarks for future reference.
It uses Google's MediaPipe Tasks Vision library. A deep learning model automatically detects body parts in the image and identifies 33 key landmark positions. This enables analysis of posture, joint angles, and body proportions.
Images showing the full body work best. Better results are achieved with good lighting, simple backgrounds, and when the person is facing forward or sideways. If multiple people are present, it detects the most central person.
Angles are calculated using three consecutive landmark points. For example, elbow angle uses the shoulder-elbow-wrist vector with the arctangent function. Results are displayed as values between 0-180 degrees.
Colors distinguish body parts: Blue for left side (left arm, left leg), red for right side (right arm, right leg), and green for central torso (shoulder and hip connections). This helps identify left-right symmetry.
Yes. You can analyze exercise postures like squats, deadlifts, and planks to verify correct joint angles. For example, check if knee angle is close to 90 degrees during squats or if the back is straight.
All processing occurs in your browser, and uploaded images are not sent to any server. After analysis, refreshing the page deletes all data. You can use it safely with privacy protection.
AI Pose Estimation is a technology that automatically detects and analyzes the positions of human body joints in images or video. It is actively used across various fields including sports science, medical rehabilitation, gaming, and AR/VR, enabling professional motion analysis with just a smartphone.
The latest pose estimation technology leverages deep learning-based convolutional neural networks (CNN). It is advancing into multi-person pose estimation (detecting multiple people simultaneously from a single image), real-time pose estimation, and 3D pose estimation that identifies body positions in 3D space.
Google's MediaPipe is a pose estimation solution that runs in real time on mobile and web. Based on the BlazePose model, it estimates 33 key full-body landmarks in 3D coordinates. Thanks to its lightweight model, real-time processing at over 30 frames per second is possible even on smartphones and web browsers.
Pose estimation is used in sports performance analysis to analyze athletes' posture and movement patterns. In medical rehabilitation, it objectively measures a patient's range of motion, while fitness apps use it to correct exercise form in real time. In the film and gaming industries, it greatly reduces motion capture costs, and it is also applied to AR fashion fitting and virtual dance lessons.