Multi-to-Multi Video Object Tracking Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video analytics systems face challenges in accurately tracking objects in complex scenarios where objects merge, split, enter, or leave the scene, leading to difficulties in maintaining accurate associations between bounding boxes and trackers.
Innovation Solution
A multi-to-multi matching process is implemented to associate multiple bounding boxes with multiple trackers, using a series of associations to minimize distances and identify unmatched entities, with the ability to handle situations like object merging, splitting, and entry/exit from the scene, by designating hidden trackers and generating new trackers as needed.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a simple one-to-one matching process is used to associate bounding boxes with trackers, then the system is computationally efficient and easy to implement, but it fails to accurately track objects in complex scenarios where objects merge, split, enter, or leave the scene
Solution Approach 1:
The matching process is segmented into multiple distinct phases: initial one-to-one matching, identification of unmatched entities, multi-to-multi matching for remaining entities, and handling of special cases (merges, splits, entries, exits). This segmentation allows the system to progressively handle increasing complexity while maintaining computational efficiency for simple cases.
Solution Approach 2:
The matching process dynamically adapts its complexity based on the scene conditions. For most frames with simple object arrangements, a basic one-to-one matching suffices. When complex scenarios are detected (unmatched trackers or bounding boxes remaining), the system dynamically transitions to multi-to-multi matching, adjusting the processing complexity to match the actual scene requirements.
2Reliability
If multiple association passes are performed to handle all matching scenarios, then tracking accuracy improves, but processing time increases
Solution Approach 1:
The system performs partial matching in the first pass, accepting that not all entities will be matched initially. This intentional incomplete matching allows the system to quickly identify unmatched entities and then apply more sophisticated multi-to-multi matching only to those specific cases, rather than applying complex algorithms to all entities in every frame.
Solution Approach 2:
The one-to-one matching is performed as a preliminary action before the more computationally intensive multi-to-multi matching. This preliminary pass pre-processes the data by establishing obvious matches and identifying unmatched entities, thereby reducing the workload for subsequent processing passes and optimizing overall processing time.
3Measurement precision
If the system uses distance minimization for matching, then association accuracy improves, but the system becomes less adaptable to objects entering or leaving the scene
Solution Approach 1:
Unmatched trackers and unmatched bounding boxes serve as intermediaries that mediate between distance-based matching and scene dynamics. By explicitly identifying and handling these unmatched entities through multi-to-multi matching, the system bridges the gap between precise distance-based association and adaptability to objects entering or leaving the scene.
Solution Approach 2:
The system changes the matching parameters from strict one-to-one distance minimization to flexible multi-to-multi association. This parameter change allows the system to consider multiple possible associations simultaneously, enabling it to adapt to dynamic scenes where objects may enter, leave, merge, or split, while still maintaining matching precision through distance considerations.
Data Source
AI summary
Techniques and systems are provided for processing video data. For example, techniques and systems are provided for matching a plurality of bounding boxes to a plurality of trackers. In some examples, a first association is performed, in which case one or more of the plurality of bounding boxes are associated with one or more of the plurality of trackers by minimizing distances between the one or more bounding boxes and the one or more trackers. A set of unmatched trackers are identified from the plurality of trackers after the first association. The set of unmatched trackers are not associated with a bounding box from the plurality of bounding boxes during the first association. A second association is then performed, in which case each of the set of unmatched trackers is associated with an associated bounding box from the plurality of bounding boxes that is within a first pre-determined distance. A set of unmatched bounding boxes is identified from the plurality of bounding boxes after the second association. The set of unmatched bounding boxes are not associated with a tracker from the plurality of trackers during the second association. A third association is then performed, in which case each of the set of unmatched bounding boxes is associated with an associated tracker from the plurality of trackers that is within a second pre-determined distance.