Sample image acquisition, cargo management method, device, equipment and storage medium

By generating a rich set of sample images through parametric virtual warehouse 3D model rendering, the problems of insufficient sample image acquisition and low annotation quality in existing technologies are solved, thereby improving the training accuracy and efficiency of image segmentation models.

CN115601547BActive Publication Date: 2026-06-09AGRICULTURAL BANK OF CHINA
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
AGRICULTURAL BANK OF CHINA
Filing Date
2022-10-28
Publication Date
2026-06-09

AI Technical Summary

Technical Problem

Existing warehouse cargo management methods suffer from insufficient sample image collection, limited content, and difficulty in ensuring annotation quality, resulting in poor training accuracy of image segmentation models.

Method used

By acquiring a parameterized 3D model of a virtual warehouse, rendering is performed using variable model parameters to generate a rendered image of the virtual warehouse scene and its labeled image. Combined with sample images of real warehouse scenes, a rich set of sample images is automatically obtained for training the image segmentation model.

Benefits of technology

It improves the richness and annotation accuracy of the sample image set, enhances the training accuracy and adjustment efficiency of the image segmentation model, and reduces the need for manual collection and annotation.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN115601547B_ABST
    Figure CN115601547B_ABST
Patent Text Reader

Abstract

The application provides a sample image acquisition, cargo management method, device, equipment and storage medium. The method comprises the following steps: acquiring a parameterized virtual warehouse three-dimensional model; at least one model parameter in the parameterized virtual warehouse three-dimensional model is variable, and the model parameters comprise: a cargo parameter in the virtual warehouse, a light source parameter entering the virtual warehouse, and a camera parameter for shooting the virtual warehouse; according to the values of the model parameters, the parameterized virtual warehouse three-dimensional model is rendered to obtain at least one virtual warehouse scene rendering image and a labeled image; according to the virtual warehouse scene rendering image and the labeled image, a sample image set is obtained; the sample image set is used for training a first preset model to obtain an image segmentation model; the image segmentation model is used for determining the area occupied by the cargo in a target warehouse scene image including a warehouse and the cargo according to the target warehouse scene image. The application improves the accuracy of image segmentation model training.
Need to check novelty before this filing date? Find Prior Art