Automated Machine Vision Workflow Synthesis Using Reinforcement Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The complexity of pattern recognition in computer vision solutions requires expert-guided algorithm selection across various stages, leading to increased effort and time in building machine vision workflows, as different workflows are needed for diverse environments and image conditions, making it challenging to synthesize effective automatic workflows.
Innovation Solution
A processor-implemented method and system for generating composable workflows in machine vision applications, which involves converting input queries into goal states, creating a problem specification file, and using a computer vision domain library to automatically generate action plans through reinforcement learning, optimizing for computational complexity, memory footprint, and performance metrics.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If expert-guided algorithm selection is used at various stages of pattern recognition, then the reliability of pattern recognition is improved, but the device complexity and time consumption increase enormously
Solution Approach 1:
The system enables automatic workflow synthesis that autonomously selects algorithms and determines processing stages without expert intervention. The synthesis system automatically analyzes image characteristics, selects appropriate algorithms from available options, and constructs workflows, making the system self-sufficient and eliminating the need for expert-guided configuration.
Solution Approach 2:
The system dynamically adjusts workflow parameters including algorithm selection, processing stages, and configuration settings based on image characteristics and task requirements. By changing these parameters automatically rather than requiring fixed expert configuration, the system achieves both reliability and reduced complexity.
2Adaptability or versatility
If multiple workflows are created for different environments and image conditions, then the adaptability is improved, but the effort and time to build new solutions increases enormously
Solution Approach 1:
The system creates a universal workflow synthesis framework that can handle multiple environments, image conditions, and task types through a single automated system. Rather than requiring separate manual workflows for each scenario, the synthesis system adapts to different conditions automatically, making the workflow building process both time-efficient and highly adaptable.
Solution Approach 2:
The system dynamically generates workflows based on real-time analysis of image characteristics and task requirements. Instead of using static pre-defined workflows, the system adapts its workflow construction dynamically, selecting and configuring algorithms and stages according to the specific conditions encountered, thereby achieving adaptability without increasing building time.
3Manufacturing precision
If manual workflow synthesis is performed for diverse image conditions and algorithms, then the manufacturing precision is improved, but the productivity decreases
Solution Approach 1:
The system replaces manual expert configuration (mechanical process) with automated algorithmic synthesis. The workflow synthesis system uses computational methods to automatically select algorithms, determine processing stages, and configure parameters, substituting human expert manual work with automated mechanical/computational processes that maintain precision while dramatically improving productivity.
Data Source
AI summary
Automation is the key to build efficient workflows with minimum effort consumption. However, there is a large gap in workflow synthesis for automated AI application development. Computer vision workflow synthesis largely rely on domain expert due to lack of generalization over solution search space for given goal. This search space for creating suitable solution(s) using available algorithms is quite vast, which makes exploratory work of solution building a time-, effort- and intellect intensive endeavor. Embodiments of the present disclosure provide system and method for goal-driven algorithm selection approach for building computer vision workflows on the fly. The system generates one or more task workflows with associated success probability depending on initial conditions and input natural language goal query by combining various image processing algorithms. Symbolic AI planning is aided by Reinforcement Learning to recommend optimal workflows that are robust and adaptive to changes in the environment.