Method and system for vision-centric deep-learning-based road situation analysis

Abstract

A method and a system for vision-centric deep-learning-based road situation analysis are provided. The method can include: receiving real-time traffic environment visual input from a camera; determining, using a ROLO engine, at least one initial region of interest from the real-time traffic environment visual input by using a CNN training method; verifying the at least one initial region of interest to determine if a detected object in the at least one initial region of interest is a candidate object to be tracked; using LSTMs to track the detected object based on the real-time traffic environment visual input, and predicting a future status of the detected object by using the CNN training method; and determining if a warning signal is to be presented to a driver of a vehicle based on the predicted future status of the detected object.

Description

technical field [0001] The present invention relates to the technical field of image processing, in particular to a vision-centered road condition analysis method and system based on deep learning. Background technique [0002] To improve the quality of mobility on the road, driver assistance systems (DAS) offer a way to highlight and enhance active and integrated safety among other things. Today, establishing advanced driver assistance systems (ADAS) to support rather than replace human drivers has become a trend in current intelligent vehicle research. These systems support drivers by enhancing their perception, providing timely warnings to avoid mistakes, and reducing the driver's control workload. [0003] ADAS systems usually use more than one type of sensor: such as image sensor, lidar, radar, etc. A single sensor cannot provide a complete, robust, accurate input. The depth perception ability of the image sensor is weak, and the resolution ability is higher than tha...

AI Technical Summary

Problems solved by technology

A single sensor cannot provide a complete, robust, accurate input

The depth perception ability of the image sensor is weak, and the resolution ability is higher than that of lidar and radar

The lateral spatial information displayed by radar is limited, because it cannot reach all, its field of view is narrow, or its resolution is reduced at long distances

Although lidar has a wide field of view and solves some of the problems mentioned above, it has other problems, such as low resolution, clustering errors, and recognition delays.

Images