Portable Video Object Tracking With Dual-AI Identity Verification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional object tracking methods in portable terminals struggle with accurately tracking a single object when multiple individuals appear on the screen or when the tracked person temporarily leaves, leading to tracking failures.
Innovation Solution
Utilizing two distinct artificial intelligence models, a first model for object identification and a second model for identity determination, to ensure that the selected object and tracking object belong to the same type, enabling effective and accurate tracking.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a single object tracking method is used, then the tracking function is simple to implement, but the tracking accuracy deteriorates when multiple individuals appear or the tracked person leaves the screen
Solution Approach 1:
The patent divides the tracking system into two separate AI models: a first model for object identification and a second model for identity determination. This segmentation allows each model to specialize in its specific function, improving overall tracking reliability while managing complexity through functional division.
Solution Approach 2:
The patent introduces an intermediary identity determination process that uses the first model's output as input to the second model. This intermediary step verifies object identity before final tracking decisions, enhancing accuracy without requiring a completely complex new system.
2Measurement precision
If multiple AI models are used for object identification and identity determination, then the tracking accuracy is improved, but the processing time and computational resources increase
Solution Approach 1:
The first AI model performs preliminary object identification and extraction before the second model conducts identity determination. This preliminary action prepares the data in advance, allowing the second model to focus specifically on identity verification, thereby reducing overall processing time despite using multiple models.
Solution Approach 2:
The patent extracts only the necessary object information from the first model's output and feeds it to the second model. By taking out only the relevant features and identifiers needed for identity determination, the system minimizes processing overhead while maintaining high accuracy.
Data Source
AI summary
According to one technical aspect of the present invention, a locked-on target based object tracking method is a method that is performed in a portable terminal and provides an object tracking function to a recorded video captured for a forward view of the portable terminal. The locked-on target based object tracking method includes setting a tracking object, identifying an object which is existing in a frame image of the recorded video using a first artificial intelligence model and setting the object as a selected object, the object and the tracking object belong to the same, and determining whether the selected object set by the first artificial intelligence model and the tracking object belong to the same object using a second artificial intelligence model trained in a different way from the first artificial intelligence model.