Multi-query Object Matching via Inverse Model Frequency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video surveillance systems for residential security struggle to effectively match multiple query samples of objects across images and videos, limiting their ability to provide accurate and timely information about activities and events.
Innovation Solution
A system utilizing Convolutional Neural Networks (CNN) for feature extraction, combined with an inverse model frequency (IMF) calculator, image scorer, model scorer, and model selector to determine confidence scores and select the most accurate model for object classification, enabling multi-query object matching.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If multiple query samples are used for object matching, then the reliability of object classification is improved, but the device complexity increases due to the need for IMF calculation, multiple scoring mechanisms, and model selection processes
Solution Approach 1:
The system segments the object matching process into distinct functional modules: an image scorer for initial matching, an IMF calculator for weight computation, a model scorer for refined scoring, and a model selector for final selection. This segmentation allows each component to specialize in a specific aspect of the matching process, improving overall reliability while making the complex system more manageable and maintainable
Solution Approach 2:
The image scorer performs preliminary filtering and scoring of query samples before they undergo the more computationally intensive IMF calculation and model scoring processes. This preliminary action reduces the burden on subsequent stages by pre-identifying promising matches, thereby improving classification reliability without proportionally increasing overall system complexity
2Measurement precision
If inverse model frequency calculation is applied to weight query samples, then the measurement precision of object matching is improved, but the loss of time increases due to additional computational steps
Solution Approach 1:
The image scorer computes preliminary scores for all query samples before the IMF calculation stage. This preliminary scoring identifies high-potential matches that can be prioritized in subsequent processing, allowing the system to achieve high measurement precision while reducing the effective processing time by focusing computational resources on the most promising candidates
Solution Approach 2:
The system applies IMF calculation and detailed model scoring selectively to query samples that meet certain criteria from the preliminary scoring stage, rather than uniformly processing all samples with the same level of detail. This partial application of computationally intensive operations maintains high measurement precision for critical matches while reducing overall processing time
Data Source
AI summary
Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for multi-query object matching based on inverse model frequency. The methods, systems, and apparatus include actions of obtaining images of a sample object, obtaining models of known objects, determining an image score for each pair of the images and the models, determining an inverse model frequency for each image based on the image scores, determining a model score for each model based on the inverse model frequencies and the image scores, and selecting a particular known object of the known objects as the sample object based on the model scores.