Hierarchical Object Recognition Loading for Memory-Limited Electronics
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing object recognition technologies using AI models inefficiently utilize memory and computational resources due to loading the entire model, which can overwhelm devices with limited memory, computational, and communication capabilities.
Innovation Solution
An electronic device selectively loads object recognition models with a hierarchical structure into volatile memory based on operation modes, using a processor to determine and load only the necessary models for identification and training, reducing computational and memory consumption.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If a touch-sensitive display is used to enable touch operations, then ease of operation is improved, but the risk of inadvertent input increases
Solution Approach 1:
The system dynamically adjusts the response behavior of the touch-sensitive display based on detected movement characteristics. When inadvertent movement is detected during touch input, the system modifies whether to register the input, making the response dynamic rather than static. This resolves the contradiction by maintaining ease of operation for intentional touches while preventing inadvertent inputs through adaptive response control.
2Ease of operation
If the device is held in one hand during video calling, then ease of operation is improved, but stability of the device composition deteriorates
Solution Approach 1:
The device performs self-diagnosis by automatically detecting its own movement characteristics during video calling using its motion sensors. The system monitors acceleration patterns and determines whether the device is being held steadily or moved inadvertently, then autonomously adjusts input registration accordingly. This self-service approach allows one-handed operation while maintaining stability through automatic detection and adjustment.
3Productivity
If touch-sensitive display operations are used, then productivity is improved, but loss of information increases due to inadvertent input
Solution Approach 1:
The system implements feedback by continuously monitoring movement characteristics during touch input and using this information to determine whether to register the input. The feedback loop detects acceleration patterns, compares them against thresholds, and adjusts input registration accordingly. This prevents inadvertent inputs that would cause information loss while maintaining productivity for intentional touches by allowing rapid input when movement characteristics indicate deliberate action.
Data Source
Figure 1A
Figure 1B~1D
Figure 1E~2
AI summary
The electronic device includes a camera, a non-volatile memory storing at least one instruction and a plurality of object recognition models, a volatile memory, and a processor, connected to the non-volatile memory, the volatile memory, and the camera, and configured to control the electronic device. The processor, by executing the at least one instruction, is configured to, based on an operation mode of the electronic device, load, to the volatile memory, a hierarchical object recognition model having a hierarchical structure corresponding to the operation mode, the hierarchical object recognition model including objection recognition models among the plurality of object recognition models, obtain information on an object by inputting an object image obtained through the camera to the hierarchical object recognition model, and determine an operation of the electronic device based on the information on the object.