Enhancing multiple object tracking accuracy via quantum annealing

Area of Science:

• Computer Vision
• Quantum Computing
• Artificial Intelligence

Background:

• Multiple Object Tracking (MOT) is crucial for image recognition but faces challenges in balancing speed and accuracy.
• Existing MOT methods struggle to meet the demands of real-time applications.

Purpose of the Study:

• To develop a novel approach for MOT that enhances both computational speed and tracking accuracy.
• To leverage quantum annealing (QA) and ensemble processing for improved MOT performance.

Main Methods:

• Utilized quantum annealing (QA) for expedited computation in MOT.
• Employed ensemble processing of object tracking methods to enhance accuracy.
• Introduced a new method for improving the matching integration process.
• Applied reverse annealing to further augment the tracking method by exploiting its sequential nature.

Main Results:

• Achieved high tracking accuracy with a minimal annealing time of 3 μs per tracking process.
• Demonstrated time-to-solution reductions exceeding 99% compared to conventional QA implementations.
• Validated the effectiveness of the proposed method through experimental testing.

Conclusions:

• The novel QA-enhanced MOT method significantly improves processing speed and accuracy.
• The approach shows great promise for real-time MOT applications.
• Potential applications include traffic flow measurement, collision prediction, and factory production management.