Recognizing Acceptance Level in Facial Expressions by Neural Network

Today, robots play an important role in the daily lives of people with disabilities and even ordinary people, so that in almost all areas of treatment and assistance, education and rehabilitation, games and entertainment, we see the presence of all kinds of social robots. Because working in the above areas requires a strong spirit and performing a specific action with constant quality many times, robots can take the place of humans well and do their job without fatigue and boredom and with a constant quality. However, one of the disadvantages that may exist in human-robot interactions is the lack of emotional mutual understanding, which means that usually the robot has no emotional understanding of human moods and sometimes this is the reason why the quality of interactions decreases. Perhaps, the perception of people's satisfaction can be considered as a major parameter in our interactions between humans and robots, meaning that creating a proper interaction always increases the level of satisfaction in humans and on the other hand, people express dissatisfaction. They can express their unwillingness to continue an interaction. Hence, this paper attemts to use a canonical neural network model to find people's level of acceptance when facing a predetermined scenario. Unlike numerous and valuable studies that use deep neural network to diagnose facial expressions and the raw image of the person as the input of the network, in this paper, the histogram vector of directional slopes of face as a characteristic vector describing the level of acceptance and a small neural network model is used as classifier. The obtained model, in addition to the high power of satisfaction, has the ability to generalize and recognize unlearnt negative emotions. Small size and low processing cost are two very important elements in the efficiency of separate systems, which are considered as two basic constraints in the model. Sometimes, other parameters are ignored to achieve these two important ones.