Lucas Arroyo Blanco
PatoOsoPatoso
1 ) Description
1 ) Requeriments
A face identification software with the ability to take samples of different users from multiple angles and train with that dataset.
- main.py to start the face identification
- take_samples.py to take samples of users from different angles
- train.py to train a model from the dataset created using take_samples.py
- OpenCV for python
- A camera plugged into the computer
- NVIDIA graphics card
- CUDA
- CudNN library
- CMake
The optional requirements improve a lot the performance of the training and of the actual execution of the program.
If you have more than one camera and want to use an other than the default one you have to modify this in order to select that one:
cap = cv2.VideoCapture(0) # Change the '0' to the device number listed by opencv
If you are unable to make the optional requirements I advise you to change a parameter in 2 scripts.
- In both train.py and main.py:
to
boxes = face_recognition.face_locations(rgb, model="cnn")
boxes = face_recognition.face_locations(rgb, model="hog")