Emotion Detection Using Deep Fusion Model

Abstract

Extracting personal emotions from a given input format such as an image has been one of the most powerful and challenging research activities in social networking. Deep learning is the trending technology in the market which provides a good working model for emotion recognition. So Deep Learning algorithms will definitely perform better than traditional methods of image processing. Automatic emotion detection can help robots communicate intelligently with humans to provide better services. To increase the accuracy of the model proposed here, we defined the emotion recognition process using the CNN fusion method, which links two deep convolutional neural networks. The overall result of this experiment is based on the dataset FER 2013 We used a fusion method consisting of two image classification models VGG 16 and ResNet 50. The feature extraction using deep residual network ResNet-50 and VGG 16 combining convolutional neural networks for facial emotion recognition. It is based on developing a detailed emotion recognition model by reading the input images to provide the recognised output according to the emotions. The developed emotion recognition model is mainly used in the healthcare industry to recognise patients’ emotions. Our model would be able to detect all the emotions of the patient, which is helpful for the doctors to treat the patient accordingly.