Emojify
- 0 Collaborators
A CNN based Facial Sentiment Analyzer and labelling tool, which uses OpenCV and Tensorflow to Put a relevant emoji on a Human face in real time ...learn more
Project status: Published/In Market
Intel Technologies
Intel Python,
MKL
Overview / Usage
This is a practical real time application of sentiment analysis where, using the frontal face of a person captured via webcam, there emotion is detected.
Just running the script loads the pretrained model, and applies it on the face captured by HaarCascade
Methodology / Approach
At the core of this project is a CNN model, trained on the FER 2013 Dataset on the Following emotional labels:
- Anger
- Fear
- Surprise
- Sad
- Happy
- Neutral
The dataset was modified so that the training data contains equal amoints of these label samples. Then the training was performed and the weights saved.
Every time the script runs, the webcam is fired up, and displays the emoji corresponding to the predicted label over the face of the person.
Technologies Used
Keras-> Tensorflow-> Intel Python
Opencv Python