How goyaltrishita/human-pose-detection-opencv works
This small Colab-oriented application detects 18 body keypoints in an image, filters them by confidence, draws skeletal connections, and displays the annotated image. The sampled Python source supports the processing stages shown; the README also documents a notebook workflow, but the notebook contents were not inspected.
Main entry points: human_pose_detection.py Load inputs · human_pose_detection.ipynb Notebook workflow
Components 13
Pose processing
- Python runner
human_pose_detection.py - Notebook workflow
human_pose_detection.ipynb - Load inputs
human_pose_detection.py - Create input blob
human_pose_detection.py - DNN inference
human_pose_detection.py - Decode keypoints
human_pose_detection.py - Draw skeleton
human_pose_detection.py - Show result
human_pose_detection.py
Inputs and output
- Input image
- TensorFlow pose model
- Annotated image
Other components
- Colab user
- OpenCV DNN
Connections 14
- Colab user → Python runner: runs
- Colab user → Notebook workflow: runs
- Python runner → Load inputs: executes
- Load inputs → Input image: reads
- Load inputs → TensorFlow pose model: loads
- Load inputs → Create input blob: passes image
- Create input blob → DNN inference: creates blob
- DNN inference → OpenCV DNN: calls
- TensorFlow pose model → OpenCV DNN: loaded by
- OpenCV DNN → DNN inference: returns heatmaps
- DNN inference → Decode keypoints: provides heatmaps
- Decode keypoints → Draw skeleton: provides points
- Draw skeleton → Show result: annotates image
- Show result → Annotated image: shows