How reiner-11/fit_vision_webcam works
Fitness Motion is a webcam exercise-tracking app: a user selects an exercise, the app captures frames, detects pose keypoints, counts completed reps, and displays angle, form cues, and FPS. Exercise-specific angle thresholds and cues are implemented separately from pose inference. The repository also supports obtaining pose-training data and evaluating trained models. Edges reflect the sampled source and README; training is documented as an external Ultralytics CLI workflow, and unsampled details are not inferred.
Main entry points: main.py Webcam app and HUD · evaluate.py Model evaluator · live_check.py Live environment check · inference.py Pose inference
Components 14
Live tracking
- Webcam app and HUD
src/main.py - Pose inference
src/inference.py
Exercise analysis
- Rep and form logic
src/exercises.py
Model data and evaluation
- Dataset downloader
download_dataset.py - Pose datasets
datasets - Pose weights
runs/pose - Model evaluator
evaluate.py - Evaluation outputs
runs/pose
Runtime checks
- Live environment check
live_check.py
Other components
- Exerciser
- Webcam
- Ultralytics YOLO
- Roboflow
- Ultralytics pose trainer
Connections 17
- Exerciser → Webcam app and HUD: selects exercise
- Webcam → Webcam app and HUD: provides frames
src/main.py - Webcam app and HUD → Pose inference: predicts and extracts
src/main.py - Pose inference → Ultralytics YOLO: runs inference
src/inference.py - Webcam app and HUD → Rep and form logic: updates state
src/main.py - Webcam app and HUD → Exerciser: shows HUD
- Dataset downloader → Roboflow: downloads from
download_dataset.py - Dataset downloader → Pose datasets: writes export
- Pose datasets → Ultralytics pose trainer: training input
- Ultralytics pose trainer → Pose weights: produces weights
- Pose weights → Webcam app and HUD: loads model
- Model evaluator → Pose datasets: evaluates split
evaluate.py - Model evaluator → Ultralytics YOLO: runs model
evaluate.py - Model evaluator → Evaluation outputs: writes reports
evaluate.py - Live environment check → Webcam app and HUD: reuses app helpers
live_check.py - Live environment check → Pose inference: checks inference
live_check.py - Live environment check → Rep and form logic: checks rep logic
live_check.py