Construction Site Entrance Vehicle Queue Length Prediction Video Dataset

Construction sites often face frequent vehicle movements and congestion, leading to inefficient resource allocation. Existing solutions mostly rely on manual management or fixed camera monitoring, which suffer from poor real-time capabilities and low detection accuracy. This dataset aims to address the issue of accurately predicting vehicle queue lengths at construction site entrances to optimize site vehicle scheduling. Video data is collected through high-definition cameras installed at the site entrance, with the collection process conducted in both clear and rainy weather, and over multiple periods to cover various scenarios. Videos are manually screened to ensure their typicality and representativeness. During annotation, videos undergo multiple rounds of review and consistency checks to ensure data consistency and accuracy. The annotation team consists of experts with over 5 years of video analysis experience. Preprocessing steps include video clipping, denoising, and keyframe extraction, with all data being structurally stored in MP4 format.