Dynamic Identification Model Selection for Moving Object Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for identifying moving objects in images, such as those used in drone air traffic control, face reduced extraction accuracy when the flight state or appearance of the target object changes, even if the object type remains the same.
Innovation Solution
An object identification apparatus and method that includes a foreground extraction unit, a state extraction unit, an identification model selection unit, and an identification unit. This system performs foreground extraction, extracts the state of each foreground based on the extraction result, selects appropriate identification models using a selection model, and identifies moving objects in the images using the selected models.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a single identification model is used for each target object type, then the device complexity is reduced, but the measurement precision deteriorates when flight state or appearance varies
Solution Approach 1:
The patent implements a dynamic identification model selection mechanism where the system automatically selects or switches between different identification models based on the extracted state of the target object (such as flight state, appearance characteristics, or environmental conditions). This dynamic adaptation allows the system to maintain high measurement precision across varying conditions without requiring a single overly complex model, thereby resolving the contradiction between device complexity and measurement precision.
2Measurement precision
If multiple identification models are selected based on extracted state, then the measurement precision is improved, but the device complexity increases
Solution Approach 1:
The patent segments the identification task by dividing it into multiple specialized identification models, each optimized for specific object states or characteristics. The state extraction unit analyzes target properties and selects the most appropriate model from the segmented set, allowing high precision for each segment while keeping individual model complexity low. This segmentation approach resolves the contradiction by distributing complexity across multiple simple models rather than concentrating it in one complex model.
Solution Approach 2:
The system changes parameters by dynamically selecting different identification models based on extracted state parameters (such as flight state, size, shape, or motion characteristics). This parameter-based model selection allows the system to adapt to varying conditions and maintain high measurement precision without permanently increasing device complexity, as the complexity is managed through conditional selection rather than simultaneous implementation of all models.
3Adaptability or versatility
If the identification model is adapted based on extracted state, then the adaptability is improved, but the processing time increases
Solution Approach 1:
The patent implements preliminary action by pre-extracting state characteristics from the target object before model selection. The state extraction unit prepares key parameters (such as flight state, appearance features, or motion patterns) in advance, which then guide the rapid selection of the appropriate identification model. This preliminary extraction of state information enables the system to adapt to different conditions without significant time penalty, as the decision-making process is streamlined by having state data ready before model selection occurs.
Data Source
AI summary
In an object identification apparatus, a foreground extraction unit performs a foreground extraction with respect to input images, and generates a foreground extraction result. A state extraction unit extracts a state for each foreground based on the foreground extraction result. An identification model selection unit selects one or more identification models based on the extracted state for each foreground by using a selection model. An identification unit identifies a moving object included in the input images using the selected identification model.