Multitask Image Inference Model with Shared Feature Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In existing multitask learning methods, task-specific feature values cannot be shared among multiple tasks, limiting the ability to utilize feature values specific to one task for estimation in other tasks.
Innovation Solution
An image processing system that includes a trained model with components to extract a common feature value from an image, task-specific feature values, and a component to concatenate these values for each task, allowing for the generation and output of inference results specific to each task.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If a common feature value is extracted from an image and task-specific feature values are extracted from this common feature value, then learning and estimation time is reduced, but task-specific feature values cannot be shared among multiple tasks
Solution Approach 1:
The model is segmented into multiple branches, each handling a specific task. The common feature value is extracted once and then distributed to multiple task-specific branches, allowing each branch to process and share feature values independently while maintaining the time efficiency of single extraction
Solution Approach 2:
Task-specific feature values are nested within the common feature value structure. Each task branch extracts and processes feature values that are contained within the common feature representation, enabling hierarchical sharing where smaller task-specific features are embedded within the larger common feature set
Data Source
AI summary
An image processing system includes a training unit that generates a trained model performing a plurality of mutually different inference tasks from an image. The trained model includes: a first component that extracts a first feature value common to the plurality of inference tasks from the image; a second component that is provided for each of the inference tasks and extracts a second feature value specific to the corresponding inference task from the first feature value; a third component that generates a third feature value by concatenating the second feature values extracted for the respective inference tasks; and a fourth component that is provided for each of the inference tasks and outputs an inference result of the corresponding inference task from the third feature value.