A Fast Feasible Region Segmentation Method Based on Asymmetric Atrous Convolution

Abstract

The invention discloses a fast feasible region segmentation method based on asymmetric atrous convolution: extracting multi-scale image features from images collected by a monocular camera, using an asymmetric atrous convolution block in the last convolution block to eliminate local noise, Obtain a distinguishable representation; the feature decoding module fuses the multi-scale image features extracted by the feature encoder point by point to obtain a high-resolution and highly discriminative image feature map; uses a classifier based on the fused image features to predict the output Feasible region segmentation results in the image scene, which divides all pixels in the image into two types: drivable area and non-drivable area. The present invention introduces a brand-new asymmetric dilated convolution module to improve the discrimination of features, greatly reduce the misjudgment of undrivable roads, and does not introduce additional calculations. Based on the lightweight module and network design, the rapid segmentation of the feasible domain is realized under the premise of ensuring accuracy.

Description

technical field [0001] The invention belongs to the technical field of computer vision, and more particularly, relates to a fast feasible domain segmentation method based on asymmetric hole convolution. Background technique [0002] In recent years, feasible domain segmentation has become a research hotspot in autonomous driving technology. Since on-board cameras have great cost advantages over high-precision 3D lidars, feasible domain segmentation with monocular images as input has become an indispensable part of autonomous driving. In the feasible domain segmentation task, the pixels in the image are predefined into three categories: background, drivable roads, and non-drivable roads (such as reverse lanes of highways, sidewalks, etc.). The feasible domain segmentation task here is to segment the drivable road pixels in the image. [0003] Early feasible domain segmentation methods used low-level features, such as color, edge, and texture, for pixel-by-pixel or block-by-...

AI Technical Summary

Problems solved by technology

[0004] Most of the existing methods based on deep convolutional networks rely on complex network design to improve accuracy, but they cause great computational complexity and are difficult to apply to real-time autonomous driving scenarios.

There are still a few methods to improve the speed of model inference by reducing the complexity of the model, but they cannot achieve real-time performance on mobile embedded platforms

Images