Training method of convolutional neural network for interpreting remote sensing images, device and apparatus

Abstract

The present invention discloses a training method of a convolutional neural network for interpreting remote sensing images, a training device of a convolutional neural network for interpreting remotesensing images and an apparatus. According to one specific embodiment of the invention, the method includes the following steps that: at least one labeled remote sensing image is determined, wherein the labeled remote sensing images contain a plurality of pixel points and labeling information corresponding to the pixel points; a first processor performs online image processing on each labeled remote sensing image, so that at least one processed labeled remote sensing image corresponding to each determined labeled remote sensing image is obtained; the first processor transmits the processed labeled remote sensing images to a second processor; and the second processor trains the convolutional neural network for interpreting the remote sensing images through the processed labeled remote sensing images. With the embodiments of the invention adopted, human and material cost and time cost for acquiring the labeled remote sensing images for training can be decreased, the efficiency of the training of the convolutional neural network for interpreting the remote sensing images can be improved, and disk space required for storing training data is decreased.

Description

technical field [0001] This application relates to the field of machine learning, specifically to the technical field of remote sensing image interpretation, and in particular to a convolutional neural network training method, device and equipment for interpreting remote sensing images. Background technique [0002] Remote sensing images have been widely used in transportation, agriculture, forestry, geology, ocean, meteorology, hydrology, military, environmental protection and other fields. In order to effectively use remote sensing images to solve practical problems, it is necessary to interpret the acquired remote sensing images to obtain the interpretation results of remote sensing images. For example, the hydrological department needs to interpret the water body area in the remote sensing image, and the transportation department needs to interpret For the road area in the remote sensing image, the forestry department needs to interpret the vegetation or forest area in t...

AI Technical Summary

Problems solved by technology

[0003] However, the data volume of remote sensing images is large, and the traditional manual interpretation method is time-consuming and laborious, and cannot guarantee the timely update of data.

Images