This paper proposes a system for searching missing persons and criminal suspects using CCTV footage. Manually identifying individuals through the numerous CCTVs installed across South Korea requires significant time and manpower. To address this, the proposed solution implements a system that leverages deep learning technology to automatically detect and identify individuals in CCTV video frames and then extract ReID (Re-Identification) features. The key feature of the system is its dynamic allocation of computing resources based on the frequency of appearances of people on each camera. By allocating more resources to cameras with higher population movement, the system improves analysis efficiency. Simulation results demonstrate that the proposed coordination algorithm significantly enhances system efficiency while maintaining a high detection rate by optimizing resource allocation.