Traffic scene semantic segmentation method based on boundary-guided context aggregation

Abstract

The invention discloses a traffic scene semantic segmentation method based on boundary-guided context aggregation. According to the implementation scheme, the traffic scene semantic segmentation method comprises the steps of 1) obtaining a data set and a segmentation label; 2) data processing; 3) constructing a segmentation model; 4) constructing a loss function; 5) training a segmentation model; and 6) segmenting the traffic scene image. According to the traffic scene segmentation model with the boundary refining module, high-level semantic boundary information is reserved while low-level contour texture information is removed, the boundary of an object can be effectively detected, context information is aggregated along the boundary of the object, the consistency of pixels of the same kind is enhanced, and the segmentation efficiency is improved. Therefore, the boundary information is effectively utilized to perform semantic segmentation on the traffic scene image. According to the method, the dependency relationship between the pixels in the boundary region and the pixels in the object can be captured, and the segmentation accuracy and robustness are effectively improved.

Description

technical field [0001] The invention belongs to the technical field of image processing, relates to automatic segmentation of traffic scene images, and can be used for automatic driving. Background technique [0002] The purpose of semantic segmentation is to assign a category label to each pixel in a given image, to achieve the classification of similar pixels, to provide rich detailed information of the image, and to have a wide range of application space and development prospects. For example, in autonomous driving scenarios, by segmenting the scene, algorithms can provide information about free space on the road, as well as information such as pedestrians and traffic signs near the vehicle. [0003] Existing segmentation methods mainly use convolution operations to expand the receptive field of view and capture the global context information. This method ignores the relationship between the interior of the object and the boundary, resulting in the loss of boundary inform...

AI Technical Summary

Problems solved by technology

Since shallow features contain not only boundary information but also texture noise inside objects, this will have a negative impact on semantic segmentation

There are also some works that use boundary information to refine the predicted results. Since the two tasks of semantic segmentation and image boundary segmentation are not orthogonal to each other, wrong boundary estimation may bring difficulties to the semantic segmentation task.

Images