Pedestrian crossing road intention recognition method based on crossing actions and traffic scene context factors

Abstract

The invention relates to a pedestrian crossing road intention recognition method based on crossing actions and traffic scene context factors. The method comprises the steps of firstly carrying out thepedestrian detection through a faster-RCNN, further searching an interested target through the motion information of a pedestrian, and extracting the motion sequence of the interested target, a surrounding traffic scene sequence and a track position, secondly, designing a three-dimensional convolutional neural network to process a motion sequence of an interested target, and acquiring behavior characteristics related to the intention of pedestrians to cross the road, then, acquiring two weights according to the elements of the local traffic scene where the pedestrian is located and the vehicle running speed, correcting the pedestrian-vehicle distance, sending the corrected distance to a multi-layer perceptron to be coded, and acquiring distance characteristics related to the road crossingintention of the pedestrian, and finally, carrying out information fusion on the behavior characteristics and the distance characteristics, carrying out dimensionality reduction on the fused characteristics by utilizing a full connection layer, and obtaining a result of whether the pedestrian crosses the road or not through softmax operation.

Description

Technical field[0001]The present invention relates to the field of intelligent transportation technology, in particular to the field of pedestrian detection and analysis, and in particular to a method for identifying pedestrians' crossing road intentions based on crossing actions and traffic scene context factors.Background technique[0002]With the continuous development of artificial intelligence, sensors and control theory, unmanned vehicles have attracted widespread attention from the academic and industrial circles, and have bright application prospects. However, unmanned vehicles also need to guarantee the rights of other road users, especially the rights of more disadvantaged pedestrians. This requires unmanned vehicles to understand the behavior of pedestrians. Among the many behaviors of pedestrians, crossing the road is the most frequent behavior and is closely related to the safety of pedestrians. Drivers can perceive whether pedestrians will cross the road through simple v...

AI Technical Summary

Problems solved by technology

[0004] Using computer vision technology, researchers can extract pedestrian outline information, shape information and historical motion information to determine whether pedestrians have the intention to cross the road, but the accuracy is not high

With the development of deep learning technology, researchers have extracted the skeleton map of pedestrians, and judged whether the pedestrian will cross according to the motion pattern of the skeleton map, which has high accuracy, but the accurate extraction of the skeleton map is limited by the distance factor

Images