Hybrid Novel Machine Learning and Computer Vision Research

With the increasing demands of society on the level of human intelligence, people are more demanding of human-like intelligent robots that can work normally in highly complex environments, perform non-specific tasks and have a high degree of initiative. This paper mainly discusses the research of hybrid novel machine learning and computer vision. This paper firstly describes the human-computer interaction based on "vision". The "visual" human-computer interaction system refers to the basic principle of human visual information, using a computer with a camera to detect and identify the user's actions through non-traditional input devices, and then perform automatic human-computer interaction in two-dimensional or three-dimensional space. Then, the machine learning prediction algorithm is introduced. Finally, the experimental design and result analysis of the robot vision learning method based on the human brain-like cognitive computing model are carried out. The experimental results show that from the final recognition rate, the recognition rate of the growing long-term memory is 93.2%, which is higher than the 91.4% of the VNAIL algorithm. After the robot has a growing long-term memory that works in conjunction with working memory, it can independently master visual cognitive ability, and incrementally store and update knowledge. Intellectual development, classification and recognition abilities are improved over methods without long-term memory.