Real Time Human Emotion Recognition Using Deep Convolutional Auto-Multiplexer

Abstract

Now a day’s one of the unsolved problem in computer vision is recognizing or understanding other people's emotions and feelings. In this paper we present a novel Deep Convolutional auto-Multiplexer (DCAM) for unsupervised feature learning. CAM is trained using conventional on-line gradient descent without additional regularization terms. A MAM-pooling layer is essential to learn biologically plausible features consistent with those found by previous approaches. Our CNN feeling recognition model achieves a performance rate of 98.91%. In this paper we have a tendency to explore the impact of coaching the initial weights in associate unsupervised manner. we have a tendency to study the result of pre-training a Deep CNN as a Convolutional Auto-Multiplexer (DCAM) in a very greedy layer-wise unsupervised fashion for emotion recognition mistreatment facial features pictures. Once trained with at random initialized weights, our CNN feeling recognition model achieves a good performance. By using this pre-trained, the performance will increase to 93.52%. Pre-training our CNN as a CAM conjointly reduces coaching time marginally.

Keywords: Emotion, MAM pooling, convolutional auto-multiplexers (CAM), convolutional neural networks (CNN)