View source for 18201329. LEARNING METHOD, RE-IDENTIFICATION APPARATUS, AND RE-IDENTIFICATION METHOD simplified abstract (TOYOTA JIDOSHA KABUSHIKI KAISHA)

=LEARNING METHOD, RE-IDENTIFICATION APPARATUS, AND RE-IDENTIFICATION METHOD=


==Organization Name==

[[:Category:TOYOTA JIDOSHA KABUSHIKI KAISHA|TOYOTA JIDOSHA KABUSHIKI KAISHA]]


[[Category:TOYOTA JIDOSHA KABUSHIKI KAISHA]]

==Inventor(s)==

[[:Category:Norimasa Kobori of Nakano-ku Tokyo-to (JP)|Norimasa Kobori of Nakano-ku Tokyo-to (JP)]][[Category:Norimasa Kobori of Nakano-ku Tokyo-to (JP)]]

[[:Category:Rajat Saini of Edogawa-ku Tokyo-to (JP)|Rajat Saini of Edogawa-ku Tokyo-to (JP)]][[Category:Rajat Saini of Edogawa-ku Tokyo-to (JP)]]

==LEARNING METHOD, RE-IDENTIFICATION APPARATUS, AND RE-IDENTIFICATION METHOD - A simplified explanation of the abstract==


This abstract first appeared for US patent application 18201329 titled 'LEARNING METHOD, RE-IDENTIFICATION APPARATUS, AND RE-IDENTIFICATION METHOD

==Simplified Explanation==

The patent application proposes a re-identification method using a machine learning model to identify a target object in image data. The method involves acquiring two sets of image data containing the target object, inputting the image data into the machine learning model to obtain output data, calculating distances between the output data in an embedding space, and determining similarity based on a threshold.

* The method uses a machine learning model to perform re-identification of a target object in image data.
* It acquires two sets of image data containing the target object.
* The image data is inputted into the machine learning model to obtain output data.
* Distances between the output data are calculated in an embedding space.
* Similarity between the target objects is determined based on a threshold and the number of distances below the threshold.

== Potential Applications ==
* Surveillance systems: The re-identification method can be used to track individuals or objects across different camera feeds, enhancing security and monitoring capabilities.
* Retail analytics: It can be applied in retail settings to track customer behavior and preferences, enabling personalized marketing and improving store layout.
* Law enforcement: The method can aid in identifying suspects or missing persons by matching images from different sources, assisting investigations.

== Problems Solved ==
* Efficient re-identification: The method provides a systematic approach to re-identify a target object in image data, reducing the manual effort and time required for identification tasks.
* Object tracking across cameras: It solves the challenge of tracking individuals or objects across different camera feeds, which can be crucial in surveillance or security scenarios.
* Overcoming variations in appearance: The machine learning model and embedding space help handle variations in lighting, pose, or occlusion, improving the accuracy of re-identification.

== Benefits ==
* Automation and efficiency: The method automates the re-identification process using a machine learning model, saving time and effort compared to manual identification.
* Improved accuracy: By utilizing an embedding space and calculating distances, the method can handle variations in appearance and improve the accuracy of re-identification.
* Scalability: The approach can be applied to large datasets and multiple target objects, making it suitable for real-world applications with high volumes of image data.


==Original Abstract Submitted==

A re-identification method for performing re-identification of a target object in image data using a machine learning model is proposed. The re-identification method comprises acquiring first image data and second image data in both of which the target object is, acquiring a plurality of first output data and a plurality of second output data by inputting the first image data and the second image data into the machine learning model, calculating a plurality of distances each of which is a distance in an embedding space between each of the plurality of first output data and each of the plurality of second output data, determining that the target object of the first image data and the target object of the second image data are similar when a predetermined number or more of the plurality of distances are less than a predetermined threshold.



[[Category:G06V40/10]]
[[Category:G06V10/77]]
[[Category:G06V10/74]]