EmotionRecognition - Show your expression (syrex)

This is a Python program that detects faces in images/webcam live feed through haarcascade and recognizes emotion using Convolutional Neural Network.

Installation

• Install required libraries via pip

  pip install -r requirements.txt

• Download syrex.weights file and put it into EmotionRecognition/weights/ folder

Usage

# emotion recognition on images
python3 detect.py \
  --image=data/test.jpg \
  --labels=data/data.labels \
  --model=cfg/syrex.json \
  --weights=weights/syrex.weights 
  
# emotion recognition on live feed from webcam
python3 detect_realtime.py \
  --labels=data/data.labels \
  --model=cfg/syrex.json \
  --weights=weights/syrex.weights

Training on your data

1. Prepare your data
   • Put your images to the EmotionRecognition/data/train/images/ folder
   • Put your labels to the EmotionRecognition/data/train/labels/ folder

       Note: filenames of images must match labels filenames

2. In order to create csv file needed for training run:

   # from EmotionRecognition/scripts/
   python3 image2pixels.py \
     --image_folder=data/train/images \
     --label_folder=data/train/labels \
     --output=data/train.csv \
     --labels_decode=data/data.labels

3. Set train parameters in EmotionRecognition/cfg/train.cfg file
4. Run training:

   python3 train.py \
     --data=data/train.csv \
     --weights_folder=weights/

License