Target Recognition and Location Based on Deep Learning

Abstract

In recent years, more and more people have invested in the research and industrialization of intelligent robots. No matter in the scene of smart home, agriculture, industry or office, the intelligent cognitive level of robot in target recognition and location is far behind that of human. Previous research have relied on computer vision, machine learning and other related technologies. Although a certain effect is achieved, the intelligence level is still low. Using deep learning algorithm and target positioning to explore the real-time and accuracy of target recognition in different environments, and then develop robotic intelligent grasping system, has becoming a research hotspot in the field of computer vision and artificial intelligence. In this paper, based on the improved deep learning algorithm-YOLO, we explored the three-dimensional positioning of the target with a monocular structured light camera. The results show that its performance in real-time, accuracy and other aspects is very significant. It effectively learns the human ability in image recognition and target location, and has the ability of rapid recognition and environmental adaptability.