How arijit2003/human-pose-estimation-using-machine-learning works
This repository is a set of standalone OpenCV image-processing examples and human-pose demos, rather than one integrated application. The main user-facing workflow is the Streamlit image app: an image is uploaded (or a demo image is used), processed with a TensorFlow-backed OpenCV pose model, and displayed with estimated keypoints. Separate scripts demonstrate webcam pose estimation with MediaPipe and static-image landmark drawing. The README is only a title, so the smaller image-processing examples are inferred from their source; no connections between standalone scripts are assumed.
Main entry points: estimation_app.py Image Pose App · video_pose_est.py Webcam Pose Demo · MediaPipe.py Static Image Demo · graph_opt.pb TensorFlow Model
Components 17
Image Processing Examples
- Color and Channels
01.py - Grayscale Save
02.py - Image Dimensions
03.py - Image Resize
04.py - Image Rotation
05.py - Shape Drawing
06.py - Text Overlay
07.py
Pose Estimation Demos
- Image Pose App
estimation_app.py - Image Input
estimation_app.py - Pose Inference
estimation_app.py - Keypoint Overlay
estimation_app.py - Webcam Pose Demo
video_pose_est.py - Static Image Demo
MediaPipe.py
Pose Runtime
- TensorFlow Model
graph_opt.pb - MediaPipe Pose
Other components
- User
- Webcam
Connections 11
- User → Image Pose App: uploads image
estimation_app.py - Image Pose App → Image Input: selects image
estimation_app.py - Image Input → Pose Inference: passes image
estimation_app.py - Pose Inference → TensorFlow Model: loads model
estimation_app.py - Pose Inference → Keypoint Overlay: returns annotated frame
estimation_app.py - Image Pose App → User: displays result
estimation_app.py - Webcam → Webcam Pose Demo: provides frames
video_pose_est.py - Webcam Pose Demo → MediaPipe Pose: processes frames
video_pose_est.py - Webcam Pose Demo → User: shows annotated video
video_pose_est.py - Static Image Demo → MediaPipe Pose: processes image
MediaPipe.py - Static Image Demo → User: shows landmarks
MediaPipe.py