Monitoring the usage and carbon impact of infrastructures such as carpooling car parks may be a challenge for local authorities. In this study, we address this issue by proposing an IoT solution that integrates both Edge processing and Cloud computing. Real-time continuous video AI processing automatically detects and stores information of car entries and exits. Then Cloud computing stores the anonymized data into a FIWARE information broker using ETSI’s NGSI-LD specification and employs query language to process and display useful statistics about carpark usage. To understand the carpooling user behaviors for each car park, as well as the various Key Performance Indicators and threshold values that direct our monitoring and alerting systems, we analyzed 1 year of data acquisition and performed in-situ surveys. Hereby, we present our results that reflect two different usages of carpooling (professional and personal) in terms of stay durations, car fidelity, car energy distributions, and carbon impact. This solution was deployed and tested on 3 car park lots, 2 of which have been permanently running for about two years.