Incentive-aware Crowdsourced Detection of Vehicles at Flee

Abstract

Law enforcement agencies (LEAs) face the persistent challenge of rapidly detecting suspicious vehicles fleeing from CCTV surveillance streams. While current machine learning (ML) and deep learning (DL) models are valuable in vehicle detection, they fall short due to the limited coverage of CCTV cameras in targeted areas, allowing a significant number of vehicles to slip through the surveillance net. A potential solution involves leveraging crowd-sourced participation, where vehicles in the specified area voluntarily contribute to detecting and reporting suspicious vehicles using advanced vehicle detection algorithms on their onboard multimedia units. However, incentivizing participant engagement remains a complex and evolving challenge in the realm of crowd sourced detection tasks. To address this multifaceted issue, the research paper titled "Incentive-aware crowd sourced detection of vehicles at flee" introduces an innovative incentive-driven solution grounded in a game theoretic approach. Adopting the Stackelberg game model, this strategic framework uses a leader-follower concept to intricately design incentives for participants, closely tied to the perceived value of the fleeing vehicle. Instead of simply encouraging vehicular nodes that might typically abstain from participation, the pro posed approach actively stimulates engagement by motivating choices that significantly contribute to the collective effort of detecting suspected vehicles in a crowd-sourced system. This approach not only enhances the efficiency of vehicle detection but also fosters a more robust and actively involved community in the fight against criminal activities on roadways.