According to foreign media reports, Perceptive Automata claims that its computer vision system can carefully observe pedestrians, interpret human body language (action), and predict pedestrian behavior. The company uses human trainers to assist its in-vehicle systems in identifying images of vehicles and other targets, and then creating machine learning algorithms. The human trainer can analyze the video of the pedestrian, and the company provides more pictures and the difficulty of identifying the target by manipulation.
Perceptive also asks the trainer, such as: “Does the pedestrian want to cross the road?†or “If you are the cyclist, will it prevent the passage of the vehicle?â€
When the picture is obscured, the human trainer will laboriously judge the body language of the pedestrian in the image, while Perceptive can track the eye activity and hesitation of the human trainer for the development of machine learning algorithms. In the given picture, it is often difficult to see the direction of the pedestrian's head. The human trainer has to prejudge according to his physical movements and think about the purpose of the pedestrian. This situation shows the role played by the human head in image recognition and pre-judgment of passers-by behavior.
According to the company, by adopting similar clues, the company quantifies the awareness and intentions of pedestrians, which will play an important role in improving the pre-judgment of pedestrian behavior by the autonomous driving system .
Perceptive Automata revealed to Wired that the company is currently working with car companies to apply for a license for the technology.
Author: Li Wenlong
test discord discord discord discord discord discord test discord discord discord discord discord discord test discord discord discord discord discord discord test discord discord discord discord discord discord test discord discord discord discord discord discord kh
test2
Bossgoo(China)Tecgnology.(Bossgoo(China)Tecgnology) , https://www.cn-gangdao.com