Using deep learning to detect smiles from a live video feed.
This project was made for detecting smile in a live video using the webcam or a pre-recorded video.
- numpy
- matplotlib
- openCV
- sklearn
- imutils
- TensorFlow/Theano
- Keras
- h5py
You can install all the required libraries by running the following command
- Filter to detect face.
- CNN for traning the model(Using LeNet architecture).
- Gather a dataset of people smiling and not smiling which is tighly cropped around the face.
- Train the Convolutional Neural Network using
train_model.py
and your custom data set(the network is designed to handle class imbalance). - Run the
detect_smile.py
to start a live video using your webcam, or you can also supply an optional argument to enter the path to a video.
Adrian Rosebrock creator of PyimageSearch