Robot Vision Learning Workflow for Automated Image Annotation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing machine learning systems in robot systems require manual annotation of images, which is time-consuming, and the startup time for machine learning is prolonged due to the need for user intervention in labeling training data.
Innovation Solution
A machine-learning device for robots that includes a program setting unit, vision execution unit, result acquisition unit, annotation unit, and learning unit, allowing for automated image capture, object detection, annotation, and machine learning using a visual sensor, thereby reducing the need for manual intervention and speeding up the training process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual annotation is performed by users to collect training data, then the quality and accuracy of labels are improved, but the time required for data preparation increases significantly
Solution Approach 1:
The system performs preliminary automated annotation using pre-trained models before final manual review, preparing draft labels in advance to reduce the time users need to spend on annotation while maintaining quality through subsequent verification
Solution Approach 2:
The system uses automated annotation models to perform annotation tasks independently, serving itself by generating initial labels without requiring immediate user intervention, thereby reducing manual workload and time consumption
2Quantity of substance
If all training data is collected and annotated before starting machine learning, then the completeness of training data is improved, but the startup time of the machine learning system increases
Solution Approach 1:
The system performs preliminary machine learning training with initially available data before all training data is collected, allowing the model to start learning early and continue improving as more data becomes available, rather than waiting for complete data collection
Solution Approach 2:
The system enables dynamic, incremental machine learning where training continues in stages as data accumulates, allowing the learning process to adapt and progress dynamically rather than requiring static complete data preparation beforehand
3Productivity
If automated annotation is used to reduce manual work, then the productivity of data preparation is improved, but the accuracy and reliability of labels may deteriorate
Solution Approach 1:
The system merges automated annotation capabilities with manual review processes, combining the speed of automated models with the accuracy of human verification to achieve both high productivity and reliable labels
Solution Approach 2:
The system implements feedback loops where automated annotations are evaluated and refined based on user corrections and performance metrics, continuously improving reliability while maintaining high annotation throughput
Data Source
AI summary
Provided is a machine-learning device which can efficiently perform machine learning. The machine-learning device comprises: a vision execution unit which captures an image of an object W by means of a visual sensor by executing a vision execution command from a robot program, and detects or determines the object W from the captured image; a result acquisition unit which acquires the detection result or the determination result for the object W by executing a result acquisition command from the robot program; an additional annotation unit which gives a label to the captured image on the basis of the detection result or the determination result for the image of the object W by executing an annotation command from the robot program, and acquires new training data; and a learning unit which performs machine learning by using the new training data by executing a learning command from the robot program.