How thetonus/face_identification_demos works
This repository is a learning and benchmarking collection of face-detection demos, not a single identification service. Each runnable script captures webcam frames, optionally resizes them, detects faces with a different backend, annotates the original frame and displays it; the MTCNN demos also report FPS. naive.py additionally defines an embedding helper, but its main loop does not call it. The README describes the project as experimental; the source excerpts show webcam input rather than general file input. Each edge below is evidenced by the cited script.
Main entry points: naive.py Naive demo loop · pytorch_mtcnn.py PyTorch demo loop · tf_mtcnn.py TensorFlow demo loop · mxnet_mtcnn.py MXNet demo loop
Components 12
Detection demos
- Naive demo loop
naive.py - CVLib detection
naive.py - PyTorch demo loop
pytorch_mtcnn.py - PyTorch MTCNN
pytorch_mtcnn.py - TensorFlow demo loop
tf_mtcnn.py - TensorFlow MTCNN
tf_mtcnn.py - MXNet demo loop
mxnet_mtcnn.py - InsightFace detector
mxnet_mtcnn.py
Face analysis
- Face embeddings
naive.py
Other components
- Demo user
- Webcam
- Video preview
Connections 16
- Demo user → Naive demo loop: runs
naive.py - Demo user → PyTorch demo loop: runs
pytorch_mtcnn.py - Demo user → TensorFlow demo loop: runs
tf_mtcnn.py - Demo user → MXNet demo loop: runs
mxnet_mtcnn.py - Naive demo loop → Webcam: reads frames
naive.py - PyTorch demo loop → Webcam: reads frames
pytorch_mtcnn.py - TensorFlow demo loop → Webcam: reads frames
tf_mtcnn.py - MXNet demo loop → Webcam: reads frames
mxnet_mtcnn.py - Naive demo loop → CVLib detection: calls
naive.py - PyTorch demo loop → PyTorch MTCNN: calls
pytorch_mtcnn.py - TensorFlow demo loop → TensorFlow MTCNN: calls
tf_mtcnn.py - MXNet demo loop → InsightFace detector: calls
mxnet_mtcnn.py - Naive demo loop → Video preview: displays annotated frames
naive.py - PyTorch demo loop → Video preview: displays annotated frames
pytorch_mtcnn.py - TensorFlow demo loop → Video preview: displays annotated frames
tf_mtcnn.py - MXNet demo loop → Video preview: displays annotated frames
mxnet_mtcnn.py