Logical Topology Inference via CPGCN Joint Optimizing With Pedestrian Re-Id

Area of Science:

• Computer Vision
• Artificial Intelligence
• Machine Learning

Background:

• Deep learning dominates pedestrian re-identification (re-id) research.
• Existing methods often overlook spatio-temporal pedestrian flow logic, relying solely on camera locations.
• Current approaches disconnect logical links between different camera views by using common object detection.

Purpose of the Study:

• To propose a novel pedestrian re-identification model incorporating logical topological inference.
• To enhance re-id accuracy by integrating multicamera logical topology and feedback mechanisms.
• To improve the understanding of pedestrian movement patterns and inter-camera relationships.

Main Methods:

• A joint optimization mechanism for pedestrian re-identification and multicamera logical topology inference.
• A dynamic spatio-temporal information driving logical topology inference using conditional probability graph convolution networks (CPGCN) and a random forest-based transition activation mechanism (RF-TAM).
• A pedestrian group cluster graph convolution network (GC-GCN) for analyzing correlations in pedestrian features.

Main Results:

• Achieved 87.3% accuracy in logical topology inference.
• Attained a top-1 accuracy of 77.4% for pedestrian re-identification.
• Reached a mean Average Precision (mAP) of 74.3% in pedestrian re-identification tasks.

Conclusions:

• The proposed model significantly improves logical topology inference and pedestrian re-identification accuracy.
• Integrating spatio-temporal logic and topological inference offers a more robust approach to re-id.
• The model demonstrates strong performance on multiple benchmark datasets and real-world scenarios.