Image training data set generation method, device and equipment and medium

Abstract

The invention discloses an image training data set generation method, device and equipment and a computer readable storage medium. The method comprises the steps: scanning a target scene, so as to obtain the three-dimensional data of the target scene; inputting the three-dimensional data into three-dimensional CG software, and constructing a three-dimensional model corresponding to the target scene by utilizing the three-dimensional CG software; performing UV mapping expansion on the three-dimensional model to obtain a virtual scene corresponding to the target scene; triggering an engine of the three-dimensional CG software, generating a plurality of images by utilizing the engine and the virtual scene, and labeling the images when the images are generated so as to obtain an image trainingdata set. According to the technical scheme, the generation efficiency of the image training data set is improved through automatic generation of the images and automatic labeling of the labels, andthe generation cost of the image training data set is reduced.

Description

technical field [0001] The present application relates to the technical field of deep learning, and more specifically, relates to a method, device, device, and computer-readable storage medium for generating an image training data set. Background technique [0002] Deep learning is a branch of machine learning. It is an algorithm that uses artificial neural networks as the framework to perform representation learning on data. When performing model calculations on image data, it is necessary to rely on image training datasets containing massive images for implementation. [0003] Currently, image training datasets are often generated manually. Specifically, use a camera or mobile phone to shoot the target scene from various angles, and shoot various positions of the target scene to obtain a large number of images, and then manually label each image with the help of tools such as labelme and labelimg to obtain an image training data set. This method can provide a more accura...

AI Technical Summary

Problems solved by technology

Specifically, use a camera or mobile phone to shoot the target scene from various angles, and shoot various positions of the target scene to obtain a large number of images, and then manually label each image with the help of tools such as labelme and labelimg to obtain an image training data set. This method can provide a more accurate image training data set for deep learning, but since a large number of images need to be taken manually and each picture needs to be manually labeled, it will reduce the generation efficiency of the image training data set and It will lead to a relatively high cost of generating image training data sets

Images