Emotion Detection

A Basic Human Emotion Detecting Algorithm that can classify a face into any of seven expression. The Data is trained using FER-2013 Dataset. The Dataset consists of Seven Different Emotions namely - Angry, Disgust, Fear, Happy, Sad, Surprise & Neutral

Requirements

• Python 3
• OpenCV
• NumPy
• Tensorflow
• Keras
• Adam
• Pandas

You can install the required dependencies by running:

pip install -r packages.txt

Installation

If you're using pycharm or any similar IDE, then you can just download and extract the zip file straight away. If not, then clone the repo using:

git clone https://github.com/S-Rawther/emote_detect.git
cd emote_detect
python train_AI.py

You can Download FER2013 Dataset CSV file from: FER2013 Once downloaded, put it in the same folder as the code.

Once the training is done, you can run the program by:

cd emote_detect
python emote_detect.py

Algorithm

The AI works by using Haar Cascade to identify face from each frame of the capture device, which is then cropped to 48 pixel square for the Neural Network. The Output is decided using the most probable emotion from the list of Softmax Score.

Additional Info

While you go through the code, you may see something called an epoch, which indicates the number of passes of the data-set the AI has completed and so it is always recommended to do multiple passes. If you wish to change the epoch value, feel free to do so. But, do remeber that more number of passes doesn't necessarily mean accurate outputs.