I have made an Emotion Detection command line tool which classifies the input image into 7 emotion categories. This classifier model is trained on the fer2013 dataset and has a model accuracy of 67.7% . I hope i'll imporve this model accuracy with time and get back with a better model. In this repository,I have included various scripts which can be used to use this classifier. The different ways are as follows:
- Capturing the input image using webcam of computer
- Taking input path of the image that you want to classify
- Real-time video input from webcam and the output is shown in real-time by bounding boxes and the emotion class written on top of it
This is a command line tool only and can be extended into a mobile application by developing a fresh mobile application and deploying the model and the code on the server.
Thanks