Analyzing the surveillence videos of an operating room and tracking the important group activities, e.g. if the members are obeying the surgical procedure or are being distracted by unexpected events, can help understanding the intraoperative situations. However, due to the complexity of the situations and the redundancy of the operation, analyze surevillence videos with least human effort still remains challenging.
In this research, we aim to automatically recognize and quantify intraoperative activities from surveillence videos of an operating room.