Generating object-based layers using object classification machine learning models

By using an object classification machine learning model, image layers are automatically generated and labeled, which solves the shortcomings of existing systems in terms of accuracy, speed and flexibility, and achieves efficient and accurate image layer generation and cross-project sharing.

CN115713125BActive Publication Date: 2026-07-17ADOBE INC
View PDF 7 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
ADOBE INC
Filing Date
2022-08-04
Publication Date
2026-07-17

AI Technical Summary

Technical Problem

Existing digital image editing systems are inadequate in terms of accuracy, speed, and flexibility. They cannot accurately generate and name image layers, require extensive user interaction, and image layers cannot be shared across projects.

Method used

Employing an object classification machine learning model, it automatically generates and labels image layers through image segmentation and object classification, automatically generates image layers and updates layer labels, and supports layer transfer and search across projects.

Benefits of technology

It improves the accuracy and efficiency of image layer generation, reduces user interaction, and enables flexible cross-project sharing and reuse of image layers.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN115713125B_ABST
    Figure CN115713125B_ABST
Patent Text Reader

Abstract

Embodiments of the present disclosure relate to generating object-based layers using object classification machine learning models. The present disclosure relates to systems, methods, and non-transitory computer-readable media for accurately, efficiently, and flexibly generating image layers and determining layer labels using machine learning methods. For example, the disclosed systems utilize image segmentation machine learning models to segment digital images and identify individual objects depicted within the digital images. Additionally, in some embodiments, the disclosed systems utilize object classification machine learning models to determine an object classification of the depicted objects. In some cases, the disclosed systems also generate image layers of digital images by generating a separate layer for each identified object (or, group of similar objects). In certain embodiments, the disclosed systems also determine a layer label for an image layer according to the object classification of the respective objects depicted in each of the image layers.
Need to check novelty before this filing date? Find Prior Art

Citation Information

Patent Citations

  • Utilizing interactive deep learning to select objects in digital visual media

    US10192129B2

  • Automatic trimap generation and image segmentation

    US10692221B2

  • Robust training of large-scale object detectors with a noisy dataset

    US11126890B2

  • Deep salient content neural networks for efficient digital object segmentation

    US20190130229A1

  • Content-aware selection

    US20200250453A1