Tracklet Clustering for Occlusion-Robust Object Tracking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing object tracking systems struggle with occlusions, especially when tracking long objects and groups of objects, leading to fragmented tracklets, computational inefficiency, and resource wastage, particularly in dynamic and densely populated scenes.
Innovation Solution
The method involves clustering tracklets based on similarity criteria and generating dynamic association graphs with adaptive edge weighting and pruning, allowing for robust tracking of long objects and groups by forming coherent tracklets and managing computational complexity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional tracking-by-detection method is used to track objects in densely populated scenes, then object detection can be performed, but tracklet fragmentation occurs due to occlusions and computational complexity increases
Solution Approach 1:
The patent merges multiple tracklets that represent the same physical object into a single unified tracklet by detecting spatial and temporal overlaps between tracklets. This combination reduces tracklet fragmentation caused by occlusions and decreases the overall number of tracklets that need to be processed, thereby reducing computational complexity while maintaining or improving tracking accuracy.
Solution Approach 2:
The patent performs preliminary tracklet merging operations during the tracking process by identifying and combining overlapping tracklets before final trajectory determination. This preliminary action prevents the accumulation of fragmented tracklets and reduces the computational burden in subsequent processing stages.
2Reliability
If multiple tracklets are created for occluded objects, then object detection coverage is improved, but resource wastage increases due to processing redundant tracklets
Solution Approach 1:
The patent combines multiple tracklets that correspond to the same physical object into a single merged tracklet by detecting overlaps in their spatial and temporal dimensions. This merging eliminates redundant processing of duplicate tracklets while preserving the detection coverage achieved through multiple tracklets during occlusion events.
Solution Approach 2:
The patent identifies and discards redundant tracklet information after merging, keeping only the essential trajectory data from the merged tracklet. This discarding of redundant information reduces computational resource consumption while the merged tracklet preserves the necessary tracking coverage.
3Reliability
If tracklet clustering is performed to reduce fragmentation, then tracking reliability improves, but additional computational processing is required
Solution Approach 1:
The patent segments the tracklet merging process into distinct phases: detecting overlaps between tracklets, determining which tracklets should be merged based on overlap criteria, and performing the actual merging operation. This segmentation allows for efficient processing by handling only the necessary merging operations rather than processing all tracklets uniformly.
Solution Approach 2:
The patent uses adjustable overlap thresholds and merging criteria that can be modified based on scene complexity and computational resources available. By changing these parameters, the system can balance between achieving high tracklet continuity and minimizing additional processing time.
Data Source
AI summary
Certain aspects of the present disclosure provide techniques for object tracking. A method generally includes detecting object(s) in a set of frames associated with a first period of time; generating first tracklets, wherein each respective first tracklet comprises a respective sequence of states associated with a respective object over the first period of time, and represents a respective first trajectory for the respective object over the first period of time; clustering two or more first tracklets into first clustered tracklet(s); and determining a second trajectory for a set of the object(s) over a second period of time based on one of the first clustered tracklets.