Location-Based AI Model Loading for Object Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Electronic devices face limitations in object recognition due to high memory and processing requirements, especially when using advanced artificial intelligence models for accurate object identification, leading to inefficiencies in user devices with limited capabilities.
Innovation Solution
An electronic apparatus is designed with a sensor, camera, and dual processors, where the first processor determines the device's location and provides area information to the second processor, which loads a specific artificial intelligence model from memory to perform object recognition, optimizing processing by using only models relevant to the current location.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If advanced artificial intelligence models are used to accurately recognize more objects, then object recognition capability is improved, but memory capacity and processing capability requirements increase significantly
Solution Approach 1:
The patent segments the object recognition task by dividing it into two distinct processing stages: a first processor handles location determination and model selection, while a second processor performs the actual object recognition using loaded AI models. This segmentation allows the system to use smaller, specialized models for specific locations rather than requiring one large universal model, thereby reducing the memory capacity needed while maintaining recognition accuracy.
Solution Approach 2:
The patent implements local quality by loading different AI models into the second processor based on the determined location of the electronic apparatus. Instead of using a single comprehensive model for all scenarios, the system selectively loads models appropriate for specific locations (e.g., indoor vs. outdoor models), optimizing memory usage by only having relevant models in memory at any given time while maintaining high recognition capability for the current context.
2Measurement precision
If advanced artificial intelligence models are used to accurately recognize more objects, then object recognition capability is improved, but processing capability requirements increase significantly
Solution Approach 1:
The patent segments processing responsibilities between two processors: the first processor performs lightweight operations (location determination and model selection), while the second processor handles intensive object recognition tasks. This segmentation allows the system to use powerful but energy-intensive AI model processing only when and where needed, rather than continuously, thereby reducing overall power consumption while maintaining high recognition capability.
Solution Approach 2:
The system applies local quality by selectively activating and loading AI models in the second processor only when the electronic apparatus is in specific locations where object recognition is needed. This avoids the power consumption associated with running comprehensive recognition models in all scenarios, optimizing processing capability usage to match actual operational needs.
3Measurement precision
If comprehensive artificial intelligence models are used for all locations, then object recognition accuracy is improved, but memory usage increases
Solution Approach 1:
The patent implements local quality by determining the electronic apparatus's location and selectively loading only the AI models relevant to that specific location into the second processor's memory. For example, if the apparatus is detected to be in an outdoor environment, only outdoor-specific recognition models are loaded, rather than loading all possible models. This maintains high recognition accuracy for the current context while significantly reducing memory usage compared to loading comprehensive models for all locations.
Solution Approach 2:
The system performs preliminary action by pre-determining the location using sensor data and map information before loading the appropriate AI models. This preliminary location determination allows the system to prepare and load only the necessary models in advance, avoiding the need to maintain all models in memory simultaneously, thereby optimizing memory usage while ensuring the right models are available for accurate recognition.
Data Source
AI summary
An electronic apparatus is disclosed. The electronic apparatus includes a sensor, a camera, a memory, a camera and a processor. The memory stores a plurality of artificial intelligence models trained to identify objects and stores information on a map. The first processor provides, to the second processor, area information on an area in which the electronic apparatus is determined, based on sensing data obtained from the sensor, to be located, from among a plurality of areas included in the map. The second processor loads at least one artificial intelligence model of the plurality of artificial intelligence models to the volatile memory based on the area information and identifies an object by inputting the image obtained through the camera to the loaded artificial intelligence model.