Dangerous Driving Behaviors (e.g., Fatigue) Audio Detection Dataset in Driver Training

In the current driver training industry, dangerous behaviors such as fatigued driving remain one of the main causes of traffic accidents. Existing fatigue detection solutions mostly rely on video monitoring or physiological signal monitoring, which have issues such as high equipment costs and privacy protection difficulties. This dataset provides a more feasible solution for dangerous behavior detection through audio data, aiming to improve detection effectiveness and coverage. Data collection is conducted using high-sensitivity microphones installed in driving simulators to gather sound data during different driving states. Multiple rounds of annotation and consistency checks are employed to ensure high data quality. The annotation team consists of over 20 experts in the traffic safety field. Data underwent preprocessing steps like noise reduction and feature extraction and is stored in WAV format, organized by driving scenarios.