A Self-Supervised Monocular Depth Estimation Method Based on Deep Learning

Abstract

The invention discloses a self-supervised monocular depth estimation method based on deep learning. The method includes: extracting the original right view I respectively r and the pyramid features of the synthesized left view, and perform horizontal correlation operations on the pyramid features to obtain multi-scale correlation features F c , and obtain the improved multi-scale correlation feature F m ; put F m Send to the visual cue prediction network in the binocular cue prediction module to generate auxiliary visual cue D r , and reconstruct the right view from the synthesized left view using the reconstructed right view and the real right view I r The image reconstruction loss between to optimize the binocular cue prediction module; the visual cue D generated by the binocular cue prediction module r Disparity Map D for Constrained Monocular Depth Estimation Network Prediction l , use a consistency loss to enhance the consistency between the two; construct an occlusion-guided constraint to assign different weights to the reconstruction errors of pixels in occluded regions and pixels in non-occluded regions.

Description

technical field [0001] The invention relates to the fields of computer vision and depth estimation, in particular to a self-supervised monocular depth estimation method based on deep learning. Background technique [0002] As one of the basic tasks of computer vision, depth perception can be widely used in the fields of autonomous driving, augmented reality, robot navigation, and 3D reconstruction. Although active sensors (eg, lidar, structured light, and time-of-flight) have been widely utilized to directly acquire scene depth, active sensor devices are usually bulky, expensive, and have high energy consumption. In contrast, the depth prediction method based on RGB (color) images has the advantages of low price and easy implementation. Among existing image-based depth estimation methods, monocular depth estimation does not rely on multiple acquisitions of the perceptual environment, and has received extensive attention from researchers. [0003] In recent years, deep lear...

AI Technical Summary

Problems solved by technology

However, existing methods usually only focus on utilizing synthesized target views to construct supervisory signals, and do not fully explore and exploit the geometric correlation between source views and synthesized target views.

Moreover, due to the occlusion between the source and target views, existing methods directly minimize the appearance difference between the synthesized target view and the real target view in disparity learning, which will lead to inaccurate predicted disparities near occluded regions.

Images