First, let's talk about human-computer interaction for vision. For a person, the most subjective thing is to look at facial expressions. Today, you can tell whether the boss is happy or not just by looking at his face. If he is happy, say a few more words; if he is not, keep a distance.
In the future, robots will also need to understand human emotions, which involves the technology of facial recognition. Of course, we also have some specific methods, including feature extraction and classification. They exist abroad as well, and we are doing them ourselves. What result was achieved after it was done? The recognition rate for the seven expressions can reach 80 percent. Of course, this kind of expression is rather exaggerated. When one is happy, they are completely different from when they are angry. But in the natural process of communication, people can't be so exaggerated. Human expressions are very subtle. When we say that someone is smiling but not smiling, it's hard to tell whether they are smiling or not. This process requires some more complex features. These techniques are needed to understand the changes in emotions in real time during future communication.
