Object Detection Model Merging Weights for Compact Deployment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing object detection devices become large-scale and complex when improving recognition performance by using multiple recognizers, making them unsuitable for deployment in video monitoring or other purposes at new sites without extensive system integration and annotation.
Innovation Solution
A method to construct a small-scale target model by merging scores from multiple object detection units using weight computation and merging units, with a target model object detection unit that corrects parameters to minimize loss based on ground truth labels, allowing for efficient deployment in new environments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple recognizers are used to improve recognition performance, then recognition accuracy is improved, but device scale and complexity increase
Solution Approach 1:
The patent merges multiple field-specific recognizers into a single target model that can be deployed at new sites. The merging unit combines scores from multiple object detection units with different weights, creating a unified model that integrates the capabilities of multiple recognizers while maintaining a compact scale suitable for deployment.
Solution Approach 2:
The target model is designed to be universal and adaptable to new sites without requiring extensive system integration. It can function as a standalone object detection device that leverages learned models from various environments, making it multi-functional across different deployment scenarios.
2Measurement precision
If multiple recognizers are deployed to improve recognition performance, then recognition accuracy is improved, but system integration and annotation requirements increase
Solution Approach 1:
The system performs preliminary learning and model training in advance using field-specific data. The target model is pre-trained with learned models from various environments before deployment, so that when deployed at new sites, it requires minimal additional integration work or annotation, having already incorporated diverse environmental knowledge.
Solution Approach 2:
The weight computation unit and merging unit act as intermediaries that automatically combine outputs from multiple object detection units. This intermediary mechanism simplifies system integration by providing a standardized way to merge results without requiring complex manual configuration or extensive annotation at the deployment site.
Data Source
AI summary
In an object detection device, a plurality of object detection units output a score indicating probability that a predetermined object exists, for each partial region set to image data inputted. The weight computation unit computes weights for merging the scores outputted by the plurality of object detection units, using weight calculation parameters, based on the image data. The merging unit merges the scores outputted by the plurality of object detection units, for each partial region, with the weights computed by the weight computation unit. The target model object detection unit configured to output a score indicating probability that the predetermined object exists, for each partial region set to the image data. The first loss computation unit computes a first loss indicating a difference of the score of the target model object detection unit from a ground truth label of the image data and the score merged by the merging unit. The first parameter correction unit corrects parameters of the target model object detection unit to reduce the first loss.