Data set expansion method and system, and storage medium

By using image data to generate models, including three-dimensional reconstruction models and style models, and generating second image data of different environmental features, the problem of inefficient and expensive traditional data collection methods under different weather and lighting conditions is solved, and efficient and low-cost data set expansion is achieved.

CN119672453APending Publication Date: 2025-03-21DITU (BEIJING) TECH CO LTD
0 Cites 0 Cited by

Patent Information

Application Number
CN202311219829.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2023-09-20
Publication Date
2025-03-21

AI Technical Summary

Technical Problem

In the fields of autonomous driving and artificial intelligence, accessing data sets suitable for a variety of situations is often a challenge, especially in different weather and lighting conditions, where traditional data collection methods are inefficient and expensive.

Method used

By acquiring the first image data of the scene captured by the image acquisition device, and using the image data to generate a model, including a three-dimensional reconstruction model and a style model, to generate the second image data. The model can generate second image data different from the environmental features in the first image data based on the reference environment characteristics, ensuring consistency of scene objects.

Benefits of technology

It realizes efficient and low-cost data set expansion, and can generate diverse data sets, covering different weather and lighting conditions, reduces data collection costs, and improves the diversity and robustness of the data set.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The embodiment of the invention discloses a data set expansion method and system and a storage medium. The method comprises the following steps: acquiring first image data of a scene acquired by image acquisition equipment; an image data generation model is obtained, the image data generation model comprises a first sub-model and a second sub-model, the first sub-model comprises a three-dimensional reconstruction model, and the second sub-model comprises a style model; and generating second image data corresponding to the first image data based on an image data generation model and the first image data, the objects in the scene presented in the first image data and the second image data being the same, and the environment features corresponding to the second image data being different from at least one of the environment features corresponding to the first image data, the environmental features include at least one of a style feature, a global illumination feature, and a direction feature.
Need to check novelty before this filing date? Find Prior Art