Object Tracking via Landmark Prediction and Periodic Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing electronic device systems for object tracking, such as facial tracking, require high computational resources due to the need for continuous object detection and tracking, which can lead to inefficiencies and increased costs.
Innovation Solution
Implementing a method that uses a controlled cascade of sequential sampling filters to track objects independently of object detection, allowing for efficient prediction and measurement of object positions using predefined landmark positions, and automatically reinitializing tracking when lost without manual intervention.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If continuous object detection is performed for tracking, then tracking reliability is improved, but computational resources and processing time increase significantly
Solution Approach 1:
The system performs object detection periodically rather than continuously. After initial detection and landmark measurement, the system uses prediction models to estimate object position in subsequent frames, only performing full detection when tracking confidence is lost or updated. This periodic detection approach maintains tracking reliability while significantly reducing computational burden.
Solution Approach 2:
The system performs preliminary actions by detecting objects and measuring landmarks in the initial frame, then uses this information to predict object positions in subsequent frames. The velocity model and displacement model are trained in advance to enable efficient prediction without requiring continuous detection, thus reducing real-time computational requirements while maintaining tracking accuracy.
2Measurement precision
If complex image processing operations are performed, then tracking accuracy is improved, but processing time and power consumption increase
Solution Approach 1:
The system extracts key landmark positions from the object detection results rather than processing the entire image. By focusing only on specific landmark points and using their positions to predict object movement, the system achieves high tracking accuracy without the computational burden of processing every pixel and complex image operations in each frame.
Solution Approach 2:
The system creates a simplified representation of the object using landmark positions and prediction models. Instead of continuously processing the full image data, the system copies essential tracking information (landmark positions) and uses mathematical models to predict future positions, reducing processing time while maintaining accuracy.
Data Source
AI summary
A method for tracking an object by an electronic device is described. The method includes detecting an object position in an initial frame to produce a detected object position. The method also includes measuring one or more landmark positions based on the detected object position or a predicted object position. The method further includes predicting the object position in a subsequent frame based on the one or more landmark positions. The method additionally includes determining whether object tracking is lost. The method also includes avoiding performing object detection for the subsequent frame in a case that object tracking is maintained.