Effective epidemic prevention is crucial for schools and offices when resuming services in light of the COVID-19 public health crisis. Designing a contactless and privacy-aware check-in process for visitor screening plays a key role in protecting venues from potential exposure to the virus; it is also important for venue managers to have timely information about the load level of the venue to make adaptive access control for visitors to lower cluster infection risk. Existing approaches either require paper-and-pen reporting of visitors to venue managers, causing human-to-human transmission risk, or collecting sensitive data about visitors to generate one-time health codes, causing serious privacy breaches. In this work, we design a contactless check-in system with anonymous visitor self-reporting, and propose an adaptive venue access control mechanism that takes into consideration both the visitor’s liability and the venue’s availability to prevent in-venue cluster infection. Specifically, our method can not only assess the visitor’s liability from their self-reported information (e.g., residence, traveling, and symptoms) using radar map analysis, but also evaluate the situation of the venue by quantifying the risk of overload against social distancing. A graph database is then employed to store check-in records between visitors and venues, which brings great performance improvement for data storage and analysis. We have deployed the proposed system in our university campus since May 2020, serving 29,791 visitors in 52 venues such as office buildings and shuttle buses. We also conduct a statistical analysis and case study to validate the performance of our system.