Automatic Passenger Counting

Automatic passenger counting (APC) aims to accurately and reliably measure ridership on public transportation, providing crucial data for optimizing service and resource allocation. Current research focuses on improving data quality through denoising algorithms (e.g., integer linear optimization) and integrating APC data with other sources like automated fare collection (AFC) systems using geostatistical models to create a unified occupancy picture across entire networks. These advancements, often employing machine learning techniques such as XGBoost, LSTMs, and self-supervised learning with convolutional neural networks, are vital for enhancing the efficiency and effectiveness of public transportation systems.