Improved siamese tracking for temporal data association

Area of Science:

• Computer Vision
• Artificial Intelligence

Background:

• Temporal image data association is crucial for accurate visual object tracking.
• The precision of data association directly impacts subsequent state estimation procedures.

Purpose of the Study:

• To propose an efficient temporal data association method using a Siamese network framework.
• To improve the accuracy and robustness of visual object tracking.

Main Methods:

• Developed an efficient approach for temporal data association.
• Integrated this approach within a Siamese network architecture.
• Employed multi-tracker collaboration to enhance performance.

Main Results:

• The proposed algorithm achieves high accuracy in temporal data association.
• The method demonstrates strong robustness in visual object tracking tasks.
• Outperformed several state-of-the-art trackers, including standard Siamese trackers.

Conclusions:

• The proposed efficient temporal data association method enhances Siamese network-based visual object tracking.
• Multi-tracker collaboration contributes to superior accuracy and robustness.
• This approach offers a significant advancement over existing tracking algorithms.