Fusing Image and Radio Frequency Data for Target Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for recognizing target objects in environments, such as board games, suffer from low recognition accuracy due to limitations in image recognition, particularly in distinguishing categories and quantities of objects.
Innovation Solution
A method that combines image recognition using neural networks with radio frequency detection to accurately determine the distribution of target objects by matching image and radio frequency recognition results, adjusting for sub-regions and confidence levels to improve accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If image recognition method is used to identify target objects, then the system can obtain position and category information, but the recognition accuracy is low
Solution Approach 1:
The patent combines image recognition technology with radio frequency recognition technology to form a fused recognition system. The image recognition module provides position and category information while the radio frequency module provides identification information, and their results are fused to achieve accurate target object distribution determination, thereby resolving the low accuracy and reliability issues of single-method recognition
Solution Approach 2:
The patent introduces a result fusion module as an intermediary that processes and integrates the recognition results from both image recognition and radio frequency recognition systems. This mediator component matches and combines the results from both methods to produce the final target object distribution information, improving overall recognition accuracy
2Measurement precision
If only image recognition is used, then the system structure remains simple, but the recognition accuracy of target object distribution is insufficient
Solution Approach 1:
The patent merges image recognition and radio frequency recognition into a unified target object distribution determination system. By combining these two recognition methods, the system achieves accurate determination of target object distribution while managing complexity through modular architecture and result fusion strategies
3Measurement precision
If conventional image recognition is used to identify game currencies, then the system can operate, but the recognition accuracy of position, category, and number is low
Solution Approach 1:
The patent applies fused recognition technology to game currency identification, combining image recognition (which provides position and visual category information) with radio frequency recognition (which provides unique identification information). This fusion compensates for the information loss in conventional image recognition and achieves accurate identification of game currency position, category, and number
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach enhances recognition accuracy by fusing image and radio frequency data, effectively identifying positions, categories, and quantities of target objects, overcoming limitations of single-sensor solutions.
Implementation Method 1
processing the image via a neural network to output the image recognition result of the target objects in the target region
Implementation Method 2
obtaining the radio frequency recognition result of the target objects in the target region via a radio frequency detection antenna arranged in the target region
Data Source
AI summary
Embodiments of the present specification provide a target recognition method, apparatus, and system. The method comprises: obtaining an image recognition result and a radio frequency recognition result of target objects in a target region, and then determining the distribution of the target objects in the target region according to the radio frequency recognition result and the image recognition result. Since the radio frequency recognition result and the image recognition result are fused, the target objects in the target region can be accurately recognized, so as to improve the recognition accuracy.