Skip to content

Latest commit

 

History

History
59 lines (50 loc) · 1.69 KB

README.md

File metadata and controls

59 lines (50 loc) · 1.69 KB

Emotion Recognition

Simple Emotion Recognition for internship project.

Requirements (What to check before you do)

Make sure you already have the following things installed:

  • Windows 8~10 or Linux Ubuntu (I'm using 18.04)
  • Numpy
  • Scipy
  • OpenCV 3 or 4
  • Scikit-learn (if you want to compare it to this library usage)
  • Keras (with Tensorflow Backend)
  • Anaconda 3
  • h5py
  • Matplotlib
  • Json
  • Webcam or Video File to do the real time programming

How to Run

Emotion Detector

  • If python 3, use python3 instead
python -c haarcascade_frontalface_default.xml \
-m checkpoints/epoch_*num*.hdf5

  • If you wish to use a video file instead of webcam add -v video_path/video_file.format at the end of the above run text

  • In my laptop, the internal webcam uses cv2.VideoCapture(0) or cv2.VideoCapture(1) while external webcam uses cv2.VideoCapture(2)

Train Recognizer

  • Run
python train_recognizer.py -c checkpoints \
-m checkpoints/epoch_*num*.hdf5 \
-s *num epoch to restart at*

or

python train_recognizer.py --checkpoints checkpoints \
--model checkpoints/epoch_*num*.hdf5 \
--start-epoch *num epoch to restart at*

Test Recognizer

  • Run
python test_recognizer.py -m checkpoints/epoch_*num*.hdf5

or

python test_recognizer.py --model checkpoints/epoch_*num*.hdf5

Acknowledgment

This project is the codification of the module "Deep Learning for Computer Vision with Python" Chapter 1~3 by Dr. Adrian Rosenberck. Great great thanks to him for making it easy for me to learn DLCV. There are some module that were not included in the module, so I kinda interpret it myself. Thank goodness it works with the remaining module.