AI Agent Retraining for Indoor Object Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Artificial intelligence models used in robot cleaners for object recognition perform poorly when the objects they were trained on differ from those in the actual indoor environment, leading to reduced recognition accuracy.
Innovation Solution
An artificial intelligence moving agent that includes a camera and a processor to photograph objects, acquire type information, receive user correction, and re-train the AI model using this correction information, allowing the model to adapt to the specific objects in the indoor space.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If an artificial intelligence model is trained using pre-collected labeling data, then the model can perform object recognition, but the recognition accuracy decreases when the trained objects differ from actual objects in the indoor environment
Solution Approach 1:
The system implements a feedback mechanism where the processor receives correction information from the user about object recognition errors. This correction information is then used to retrain the artificial intelligence model, creating a closed-loop system that continuously improves recognition accuracy based on actual performance deviations.
Solution Approach 2:
The system performs preliminary object recognition using the pre-trained artificial intelligence model before receiving user feedback. This allows the system to have an initial recognition capability that can be subsequently refined, rather than waiting for user input from the start.
2Measurement precision
If the artificial intelligence model is retrained using user correction information, then the recognition accuracy for specific objects improves, but the system complexity increases
Solution Approach 1:
The system performs self-improvement by automatically retraining its own artificial intelligence model using correction information received from the user. The processor itself carries out the retraining operation, making the system self-adapting without requiring external intervention or complex infrastructure.
Solution Approach 2:
The processor serves multiple functions: it performs initial object recognition, receives and processes user correction information, and executes model retraining. This multi-functionality reduces the need for separate dedicated components for each task, thereby limiting the increase in system complexity.
Data Source
AI summary
An artificial intelligence moving agent is provided. The artificial intelligence moving agent includes: a camera configured to photograph an image, and a processor configured to photograph an object, acquire type information of the object by providing an image of the photographed object to an artificial intelligence model, acquire correction type information designated by a user with respect to the image of the photographed object, and train the artificial intelligence model by using the correction type information.