Systems & Technology Research (STR) proposes to develop Vehicle Reidentification-Aided Network Topology Inference (VRANTI), a novel system for estimating proximity network graphs of traffic cameras to facilitate intelligence applications such as tracking and monitoring of traffic systems. Network inference will be performed using statistical analyses of features extracted from camera video feeds, and wil not rely on any geospatial location information for the cameras. We will develop our algorithms using both real-world data and high-fidelity video simulation environments. In Phase I, we will (1) construct realistic, large-scale simulated traffic camera datasets, (2) implement deep learning computer vision models to extract features from video data, (3) perform monocular camera pose estimation, and (4) develop a topology inference algorithm to induce network graphs. In Phase II, we will refine and implement joint optimizations in our models to increase inference speed and reidentification accuracy. With this design we will extend capabilities to map large scale camera networks to real world positions and orientation to enable automated end to end camera network inference.