Fighting and Brawling Action Recognition Dataset

The current security industry is facing increasingly complex security risks, especially in public places such as campuses and commercial areas, where fighting incidents occur from time to time. Existing surveillance systems lack sufficient reaction speed to emergencies and efficient automatic recognition means. This dataset aims to solve the problem of automatic detection of fighting behaviors in real-time monitoring, improving the response efficiency of security personnel. The dataset contains diverse fighting scenarios sourced from environments like campuses, commercial areas, and bars, recorded using high-definition camera equipment to ensure video quality. During data collection, we strictly executed multiple rounds of annotation and consistency checks to ensure the accuracy and reliability of the annotations. The data is stored in MP4 format to facilitate subsequent processing and analysis.