Neural network search method and device for binocular vision matching

Abstract

The present invention provides a neural network search method and equipment for binocular vision matching. The method includes: acquiring training data, the training data including binocular images and corresponding parallax data; using the training data at the unit level Search the structure of the feature extraction network and the three-dimensional matching network in the binocular vision matching model in the search space and the network-level search space; optimize the difference between the parallax data obtained in the search process and the parallax data in the training data. a set of weight parameters, said second set of weight parameters, and network weights until the search process converges; retaining at least some of said operations according to the value of the first set of weight parameters, and retaining at least one path according to the value of the second set of weight parameters, and Optimized feature extraction network and 3D matching network based on preserved operations and paths.

Description

technical field [0001] The invention relates to the technical field of neural network search, in particular to a neural network search method and device for binocular vision matching. Background technique [0002] Humans have the ability to observe, reconstruct and understand the three-dimensional world through binoculars. Accurate perception and reconstruction of scenes is crucial for human decision-making. Binocular vision matching (Stereo matching) is committed to giving computers the ability similar to human eyes, and its research goal is to calculate the disparity map (disparity map) from the two-dimensional color images obtained by the binocular camera. After obtaining the disparity information, the depth information and 3D information of the original image can be easily obtained according to the projection model, so this technology has a wide range of applications in the fields of 3D scene reconstruction, robotics, and automatic driving. Binocular vision matching is ...

AI Technical Summary

Problems solved by technology

[0005] In addition, in the binocular vision matching method based on deep learning, the cost of 3D matching needs to repeatedly use a large number of floating-point operations, and cascade layers must be used to instantiate these calculations to ensure the effect, resulting in a large number of calculations. power consumption

Images