Depth prediction method based on binocular vision and related equipment

Abstract

The invention provides a depth prediction method based on binocular vision and related equipment, and the method comprises the steps: obtaining an image of a binocular camera, and obtaining a first-eye image and a second-eye image; determining an initial parallax according to the first eye image and the second eye image; inputting the first eye image, the second eye image and the initial parallax into a trained parallax network model to obtain a target parallax; and determining the target depth according to the target parallax. Through the initial parallax, the calculation resources required in the model processing process can be reduced, and the calculation amount is reduced.

Images