Lane line detection method, device and equipment and computer readable storage medium

Abstract

The embodiment of the invention discloses a lane line detection method, device and equipment and a computer readable storage medium. The method comprises the following steps: acquiring a front view in a driving scene; performing multi-scale feature extraction on the front view to obtain front view features; performing lane line prediction on the front view features by using a preset lane line detection model to obtain offsets of multiple groups of feature points, visibility of the multiple groups of feature points and confidence of the multiple groups of feature points; determining a target lane line according to the offset of the plurality of groups of feature points, the visibility of the plurality of groups of feature points, the confidence of the plurality of groups of feature points and a plurality of preset anchor lines; wherein the plurality of preset anchor point lines comprise a plurality of groups of anchor point lines with different starting points, and the inclination angles of the anchor point lines with the same starting point are different. The lane line is determined according to the offset, the visibility, the confidence coefficient and the anchor point line mode by setting the appropriately dense anchor point lines with multiple inclination angles, so that the lane line detection method can better adapt to a large-curvature curve, and the lane line detection accuracy is improved.

Description

technical field [0001] The present application relates to the field of computer technology, and in particular, to a lane line detection method, apparatus, device, and computer-readable storage medium. Background technique [0002] With the rapid development of artificial intelligence, autonomous driving is possible. In order to ensure the safety of autonomous vehicles, autonomous vehicles usually need to have the function of lane line detection. Accurate detection of lane lines is an important prerequisite for functions such as lane departure warning, lane keeping, and lane change. [0003] The detection scheme of the lane line in the prior art is to perform semantic segmentation on the input image (by classifying the pixels into two categories to determine whether the pixel belongs to the lane line or the background), to represent the pixel by an embedded vector, and to perform clustering according to the semantic segmentation result and the embedded vector. class to get l...

AI Technical Summary

Problems solved by technology

[0004] However, the above-mentioned existing technologies cannot be adapted to complex lane line scenarios, such as large curvature curves and side roads, etc., which reduces the accuracy of lane line detection

Images