Object Re-Identification Image Updates for Accuracy and Speed
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing re-identification systems face accuracy issues when multiple registered object images are stored for a known object, leading to decreased processing speed due to increased re-identification times without significant improvement in accuracy.
Innovation Solution
An image processing apparatus that includes a registered object image storage unit, an unknown object image acquisition unit, and a registered object image update unit, which adds new object images to the storage unit only when all existing images fail to match, maintaining variation without unnecessary increase in image count.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple registered object images are stored for one known object, then re-identification accuracy is improved, but processing speed is lowered due to increased number of re-identification comparisons
Solution Approach 1:
The system performs preliminary actions by storing multiple registered object images in advance for each known object. When re-identification is needed, the system can quickly compare the unknown object image against these pre-stored images, improving accuracy without requiring real-time image capture and processing. The multiple registered images are prepared beforehand to cover various poses, lighting conditions, and angles of the same object.
Solution Approach 2:
The system changes parameters by varying the number of registered object images stored for each known object. Instead of using a fixed single image or a fixed large number of images, the system dynamically adjusts the number of registered images based on the specific object and requirements. This allows optimization between accuracy and processing speed by selecting an appropriate number of registered images for each object type.
2Productivity
If a single registered object image is stored for one known object, then processing speed is maintained, but re-identification accuracy deteriorates when the unknown object image does not match the single registered image
Solution Approach 1:
The system performs preliminary actions by storing multiple registered object images in advance for each known object. When re-identification is needed, the system can quickly compare the unknown object image against these pre-stored images, improving accuracy without requiring real-time image capture and processing. The multiple registered images are prepared beforehand to cover various poses, lighting conditions, and angles of the same object.
Solution Approach 2:
The system changes parameters by varying the number of registered object images stored for each known object. Instead of using a fixed single image or a fixed large number of images, the system dynamically adjusts the number of registered images based on the specific object and requirements. This allows optimization between accuracy and processing speed by selecting an appropriate number of registered images for each object type.
Data Source
AI summary
A registered object image storage unit stores at least one image of an object as a registered object image in association with an object identification number that uniquely identifies the object. The unknown object image acquisition unit acquires an unknown object image that is an image of an unknown object. An object re-identification unit re-identifies the acquired unknown object image by using the registered object image stored in the registered object image storage unit. When a condition that all registered object images stored in the registered object image storage unit in association with the object identification number of the object in the unknown object image whose re-identification succeeded differ from the unknown object image, a registered object image update unit adds the unknown object image to the registered object image storage unit as a new registered object image in association with the object identification number.


