Segmentation Neural Network for Accurate Object Selection

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional digital object selection systems suffer from inaccuracies, inefficiencies, and inflexibility in generating object segmentations from digital images, requiring excessive user interactions and processing power, and often produce imprecise results.

Innovation Solution

The system employs a deep neural network to process object user indicators and initial object segmentations, using a salient object neural network to generate refined and accurate object segmentations by combining user inputs with the digital image, reducing the need for manual tracing and improving computational efficiency.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If conventional object selection systems use predictive methods to automatically generate object selection, then productivity is improved, but measurement precision deteriorates

Engineering Contradiction:

Improveobject selection speedVSAvoidobject segmentation accuracy

Core Design Contradiction:

ProductivityVSMeasurement precision

Solution Approach 1:

The system implements feedback by detecting user corrections to automatically generated segmentations and using this feedback to retrain the predictive model. The neural network learns from user interactions, continuously improving accuracy while maintaining automated operation.

Inventive Principle:

Principle #23Feedback

Solution Approach 2:

The system performs self-service by automatically generating initial segmentations without user intervention, then using detected user corrections to self-improve its predictive capabilities through model retraining, reducing the need for manual tracing over time.

Inventive Principle:

Principle #25Self-service

2Measurement precision

If conventional systems require user tracing to select objects, then measurement precision is improved, but productivity deteriorates

Engineering Contradiction:

Improveobject selection accuracyVSAvoidobject selection efficiency

Core Design Contradiction:

Measurement precisionVSProductivity

Solution Approach 1:

The system performs preliminary action by generating an initial object segmentation automatically before user interaction. This preliminary segmentation provides a head start, reducing the amount of user tracing needed while maintaining accuracy through subsequent user corrections.

Inventive Principle:

Principle #10Preliminary action

3Measurement precision

If deep neural network processing is implemented, then measurement precision is improved, but use of energy deteriorates

Engineering Contradiction:

Improveobject segmentation accuracyVSAvoidcomputational processing power

Core Design Contradiction:

Measurement precisionVSUse of energy by moving object

Solution Approach 1:

The system applies partial action by using the deep neural network only when necessary - specifically for generating initial segmentations and processing user corrections - rather than continuously. This selective use of computational power maintains accuracy while reducing overall energy consumption.

Inventive Principle:

Principle #16Partial or excessive action

4Ease of operation

If conventional systems implement automatic object selection, then ease of operation is improved, but adaptability deteriorates

Engineering Contradiction:

Improveuser interaction requirementVSAvoidhandling various user inputs

Core Design Contradiction:

Ease of operationVSAdaptability or versatility

Solution Approach 1:

The system implements dynamics by making the object selection process adaptive and flexible. It can operate in fully automated mode for simple cases or accept various types of user inputs (tracing, clicking, erasing) for more complex scenarios, adjusting its behavior based on the specific situation and user preferences.

Inventive Principle:

Principle #15Dynamics

Data Source

PatentUS11676279B2Utilizing a segmentation neural network to process initial object segmentations and object user indicators within a digital image to generate improved object segmentations

Publication Date: 2023.06.13 ADOBE INC

AI summary

The present disclosure relates to systems, non-transitory computer-readable media, and methods that utilize a deep neural network to process object user indicators and an initial object segmentation from a digital image to efficiently and flexibly generate accurate object segmentations. In particular, the disclosed systems can determine an initial object segmentation for the digital image (e.g., utilizing an object segmentation model or interactive selection processes). In addition, the disclosed systems can identify an object user indicator for correcting the initial object segmentation and generate a distance map reflecting distances between pixels of the digital image and the object user indicator. The disclosed systems can generate an image-interaction-segmentation triplet by combining the digital image, the initial object segmentation, and the distance map. By processing the image-interaction-segmentation triplet utilizing the segmentation neural network, the disclosed systems can provide an updated object segmentation for display to a client device.