GAN-Based Synthetic Data for Analog Gauge Digitization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for digitizing analog gauges in industrial settings are inefficient, requiring significant per-gauge customization and manual data collection, which is costly and resource-intensive, especially when gauges operate in out-of-normal regions, posing safety and regulatory risks.
Innovation Solution
A computer vision system that uses a geometric model and a generative adversarial network (GAN) to generate synthetic training data from a few actual gauge images, allowing for the training of a machine learning model to predict gauge readings and automate digitization with minimal human effort.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If manual data collection by physically manipulating the indicator position is used, then training data can be gathered, but the process is tedious and infeasible in many cases
Solution Approach 1:
The patent uses image synthesis technology to create virtual copies of gauge images with indicators at various positions. Instead of physically manipulating the indicator to capture multiple images, the system generates synthetic images that replicate the gauge appearance with indicators positioned throughout the entire range, thereby obtaining comprehensive training data without manual intervention
Solution Approach 2:
The system performs preliminary actions by pre-processing gauge images to extract geometric models and characteristics before generating synthetic training data. This includes identifying gauge face features, scale markings, and indicator properties in advance, which then enables automated generation of training images across all possible indicator positions without requiring physical manipulation during data collection
2Quantity of substance
If the gauge is operated to gather training images in out-of-normal operation region, then comprehensive training data can be obtained, but it may damage the machine or violate procedures/regulations/laws
Solution Approach 1:
The patent creates synthetic copies of gauge images with indicators at all possible positions, including out-of-normal regions, without actually operating the physical gauge in those conditions. The image synthesis system generates virtual representations of the gauge face with indicators positioned throughout the entire range, thereby obtaining comprehensive training data while avoiding any physical manipulation that could damage the machine or violate safety regulations
Solution Approach 2:
The system introduces an intermediary layer of image synthesis technology between the desired training data and the physical gauge. Instead of directly manipulating the physical gauge indicator to capture images at all positions (which could cause damage), the synthesis system acts as an intermediary that creates virtual images representing all possible indicator positions, including hazardous regions, without physical contact with the actual gauge mechanism
3Quantity of substance
If the gauge goes through its entire normal operation region to gather training images, then comprehensive training data can be collected, but it takes a long time and requires large resources
Solution Approach 1:
The patent uses image synthesis to create virtual copies of gauge images with indicators at various positions simultaneously, rather than sequentially capturing images by physically moving the indicator through the entire range. This allows the system to generate comprehensive training data representing all indicator positions in a single operation, dramatically improving data collection efficiency and reducing resource requirements
Solution Approach 2:
The system performs preliminary extraction of geometric models and gauge characteristics from a single or few actual gauge images. This pre-processing step captures all necessary information about the gauge face, scale markings, and indicator properties, which then enables automated generation of comprehensive training images without requiring the gauge to physically traverse its entire operating range, thereby significantly reducing the time and resources needed for data collection
Data Source
AI summary
A system for analog gauge monitoring uses a machine learning model for computer vision that is trained using synthetic training data generated based on one or a few images of the gauge being monitored and a geometric model describing the scale and the indicator of the gauge. In some embodiments, the synthetic training data is generated using an image model implemented as a generative adversarial network (GAN) type neural network and trained to modify an image of a given gauge such that the gauge face is preserved while the gauge indicator is added to or removed from the image of the given gauge for any given gauge.