Intermittent Object Detection and Tracking for Mobile Video
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing object detection techniques in video images are computationally intensive and power-hungry, making them unsuitable for real-time camera control in mobile devices, and often result in unreliable object tracking and increased latency.
Innovation Solution
Implementing an efficient object detection and tracking system that operates intermittently on a subset of frames, using object detection only on predetermined frames and tracking on the remaining frames, reducing processing resources and latency, and assigning unique identifiers to detected objects for reliable tracking.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If object detection is performed on every frame of video, then object detection reliability is improved, but power consumption and computational load increase significantly
Solution Approach 1:
The system performs object detection periodically on selected frames rather than continuously on every frame. The frame selector identifies specific frames based on temporal intervals or detection confidence thresholds, reducing the frequency of computationally intensive detection operations while maintaining reliable object identification through tracking on intermediate frames.
Solution Approach 2:
The video processing pipeline is segmented into two distinct stages: detection stages where full object detection is performed on selected frames, and tracking stages where lighter-weight tracking algorithms operate on all frames. This segmentation allows the system to balance detection reliability with power consumption by applying heavy computation only when necessary.
2Measurement precision
If object detection is performed on every frame, then object location accuracy is improved, but latency increases
Solution Approach 1:
The system performs preliminary object detection on selected frames to establish initial object locations and identities. Subsequent tracking operations use these preliminary detections to predict object positions in between detection frames, reducing the need for repeated full detection operations and thereby lowering latency while maintaining location accuracy.
Solution Approach 2:
A tracking algorithm acts as an intermediary between periodic detection operations. The tracker maintains object state information and predicts object locations during intervals between detection frames, allowing the system to achieve smooth, accurate object location tracking without performing computationally intensive detection on every frame, thus reducing latency.
3Reliability
If continuous object detection is performed, then object tracking reliability is improved, but processing resource consumption increases
Solution Approach 1:
The system implements periodic detection where full object detection is performed only on selected frames rather than continuously. Between detection frames, the system relies on tracking algorithms to maintain object identities and locations. This periodic approach reduces processing resource consumption by approximately 65% while maintaining tracking reliability through confidence-based validation and detection updates.
4Measurement precision
If object detection is performed on every frame, then detection accuracy is improved, but power-hungry computation increases
Solution Approach 1:
The system performs detection accurately on periodically selected frames rather than continuously on every frame. The frame selector identifies optimal detection frames based on temporal intervals or scene change detection, ensuring that detection accuracy is maintained at critical moments while reducing overall computational energy consumption by eliminating redundant detection operations on frames where objects are stable and well-tracked.
Data Source
AI summary
Techniques described herein provide efficient object detection and tracking in video images, such as may be used for real-time camera control in power-limited mobile image capture devices. The techniques include performing object detection on a first subset of frames of an input video, detecting an object and object location in a first detection frame of the first subset of frames, and tracking the detected object on a second subset of frames of the input video after the first detection frame, wherein the second subset does not include frames of the first subset.