Street Scene is a dataset for video anomaly detection. The dataset includes 46 training and 35 test videos captured from a USB camera. The videos include daytime overhead views of two-lane streets with bike lanes and sidewalks, and various activity scenes such as cars driving, turning, etc. The dataset has a total of 56,847 frames for training and 146,410 frames for testing. The dataset contains a total of 205 naturally occurring anomaly events (jaywalking, illegal U-turns, etc.).