Gaze-Guided Object Tracking for Reliable Target Re-Identification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing mobile device object tracking systems require physical user interaction for target object selection and suffer from reliability and robustness issues, such as target loss and low confidence scores.
Innovation Solution
Integrate an object tracker and an eye tracker in mobile devices to utilize gaze estimation for automatic target object selection, verification, and re-identification, enabling real-time user intention detection without physical interaction, and enhancing reliability and robustness.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If physical user interaction is required for target object selection, then user intention can be detected, but user interaction complexity increases and operation convenience decreases
Solution Approach 1:
The patent replaces physical interaction mechanisms (touchscreen clicks, button presses) with an optical-based gaze estimation system. The eye tracker captures eye movement data and processes it to determine user intent, substituting mechanical user-device interaction with an optical sensing and computational approach. This eliminates the need for physical contact while maintaining accurate intention detection.
Solution Approach 2:
The patent introduces gaze estimation as an intermediary mechanism between the user and the object tracking system. Instead of direct physical interaction, the user's gaze direction serves as an indirect signal that the system interprets to identify the intended target object. This intermediary layer enables natural, contactless interaction while preserving intent detection accuracy.
2Reliability
If traditional object tracking is used without gaze estimation, then device complexity remains low, but tracking reliability and robustness decrease due to target loss
Solution Approach 1:
The patent merges the eye tracker module with the existing object tracking system. The gaze estimation functionality is integrated into the tracking pipeline, allowing the system to use gaze information to initialize, verify, and recover target tracking. This combination enhances reliability without requiring completely separate systems, as the gaze module works synergistically with the object tracker.
Solution Approach 2:
The patent applies preliminary action by using gaze estimation to predict and prepare for target selection before traditional detection methods are fully executed. The system uses gaze direction to pre-identify potential target objects, allowing the object tracker to focus computational resources on verifying and maintaining tracking of gaze-indicated targets, thereby improving reliability while managing complexity.
3Extent of automation
If gaze estimation is integrated into object tracking, then automatic target selection and re-identification improve, but computational requirements and processing time increase
Solution Approach 1:
The patent uses gaze estimation to perform preliminary target identification before the object tracking algorithm fully executes. By determining the user's intended target in advance through eye movement analysis, the system can initialize tracking more efficiently and reduce the computational search space, thereby minimizing additional processing time while maximizing automation.
Solution Approach 2:
The patent implements a feedback loop where gaze estimation continuously provides real-time information about user intent to the object tracker. This ongoing feedback allows the system to dynamically adjust tracking priorities and verify target selection without requiring complete re-processing, thus maintaining high automation levels while controlling processing time through iterative refinement rather than exhaustive computation.
Data Source
AI summary
Mobile devices such as smartphones, comprising a first camera having a first field of view FOV1 pointed towards a scene to be photographed, the first camera configured to capture first image data, a second camera having a second field of view FOV2 pointed towards a scene that includes eyes of a user, the second camera configured to capture second image data, and a processor including an object tracker and an eye tracker configured to use the first image data to perform object tracking and to use the second image data to perform gaze estimation, wherein the object tracker is configured to use gaze estimation to perform an action selected from the group consisting of selection of an object in the scene that is to be tracked by the object tracker, verification of a tracked object in the scene and re-identification of a tracked object in the scene.