How vishl9455/facemask_detector works
This repository provides face-mask detection through a FastAPI image-upload interface and a separate OpenCV webcam program. Both detect faces with YOLO, preprocess each face, and classify it with a VGG16-based mask model. The training notebook is a separate model-development path. The API and webcam workflows are visible in sampled source; training details and frontend behavior are based on the README because those files were not sampled.
Main entry points: app.py Prediction API · webcam.py Webcam Detection · model.keras Classifier Model · index.html Web Frontend
Components 16
Web Interface
- Browser
- Web Frontend
static/index.html - Prediction API
app.py
Detection and Classification
- Image Decode
app.py - YOLO Face Detector
app.py - Face Crop
app.py - Resize and Normalize
app.py - Mask Classifier
app.py - Mask Result
app.py - Classifier Model
model.keras - YOLO Model
model.pt - Webcam Detection
webcam.py - Annotated Video
webcam.py
Model Training
- Classifier Training
train_model.ipynb
Other components
- User
- Webcam
Connections 16
- User → Browser: uses
- Browser → Web Frontend: loads
- Web Frontend → Prediction API: uploads image
- Prediction API → Image Decode: decodes upload
app.py - Image Decode → YOLO Face Detector: detects faces
app.py - YOLO Face Detector → Face Crop: provides boxes
app.py - Face Crop → Resize and Normalize: crops faces
app.py - Resize and Normalize → Mask Classifier: prepares input
app.py - Mask Classifier → Mask Result: labels prediction
app.py - Prediction API → Classifier Model: loads
app.py - Prediction API → YOLO Model: loads
app.py - Webcam → Webcam Detection: streams frames
webcam.py - Webcam Detection → YOLO Model: detects faces
webcam.py - Webcam Detection → Classifier Model: classifies crops
webcam.py - Webcam Detection → Annotated Video: annotates frames
webcam.py - Classifier Training → Classifier Model: trains and saves
README.md