MULTI-CAMERA OBJECT TRACKING WITH A TWO-STAGE GRAPH MODEL

Summary from the TLDR AI newsletter

This study introduces a new graph model for Multi-Camera Multi-Object Tracking (MC-MOT) that focuses on both space and time. Unlike previous methods that often make mistakes, this model first connects objects across different cameras and then uses that info over time, making tracking more accurate and ready for real-world use.