Object Tracking via Probability-Based Trajectory Association
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Multitarget tracking of pedestrians and sports players in videos is challenging due to clutter and occlusions, leading to frequent missed or false detections and incorrect associations of similar-looking objects.
Innovation Solution
A system and method for tracking objects in videos using a motion tracking application that detects objects in video frames by generating background reference frames, foreground masks, and occupancy maps, and computes optimal trajectories using probabilities and algorithms like maximum flow or Hungarian algorithms to associate detections across frames.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional object detection and tracking algorithms are used in crowded scenes with similar-looking objects, then the system can process videos in real-time, but the detection accuracy and association reliability deteriorate due to clutter and occlusions
Solution Approach 1:
The patent segments the tracking problem into multiple independent components: detection stage (identifying objects in each frame), association stage (linking detections across frames using data association algorithms), and trajectory computation stage (calculating optimal paths). This segmentation allows each component to be optimized independently, improving overall detection accuracy without proportionally increasing system complexity.
Solution Approach 2:
The patent introduces intermediate representations such as detection boxes, confidence scores, and association probabilities as mediators between the raw video input and final trajectories. These intermediaries enable sophisticated processing (handling occlusions and similar objects) without requiring the entire system to become overly complex, as each intermediary carries specific information needed for its stage.
2Reliability
If advanced algorithms like data association and trajectory optimization are implemented to improve tracking accuracy, then association reliability improves, but computational complexity and processing time increase
Solution Approach 1:
The patent performs preliminary actions by pre-computing detection results for all objects in each frame before performing association. Detection boxes and confidence scores are calculated in advance, allowing the data association algorithm to work with pre-processed data rather than raw pixels. This reduces the computational burden during the time-critical association and trajectory computation stages, maintaining reliability while reducing processing time.
Solution Approach 2:
The patent implements dynamic trajectory computation where the optimal trajectory is recalculated based on current detections and previous trajectories. The system adapts to changing scene conditions (occlusions, new objects entering) by dynamically updating associations and recomputing trajectories only when necessary, rather than using static pre-computed paths. This maintains high association reliability while minimizing unnecessary computational overhead.
Data Source
AI summary
There is provided a system for tracking objects. The system includes a processor and a memory for storing a plurality of sensory data frames. The processor determines a first hypothesized location for each of the objects in each of the plurality of sensory data frames. For each of the plurality of sensory data frames, the processor determines probabilities that the first hypothesized location of each of the objects in a sensory data frame of the plurality of sensory data frames is the same as the first hypothesized location of another object in an adjacent sensory data frame. The processor computes a first optimal trajectory for each of the objects using an algorithm based on the probabilities, checks the first optimal trajectory for each of the objects, and accepts or rejects the first optimal trajectory for each of the objects.