Multi-Person Detection of Drivers Based on Yolo Network

Subway train drivers abide by the operations requirements to routinely check a myriad of system parameters and indicators to ensure safe operation. It is important to ensure that the driver have correctly performed the entire set of routine operations without omission. It is therefore hoped that introducing real-time monitoring to the on-board surveillance system can replace human efforts in favor for improved safety on the driver’s side. In this paper we investigate the objective detection methods to accomplish open pose estimation. We take a good method in doing such task as it satisfies all the requirements: real-time, high accuracy, works for both RGB and greyscale input, multi-person detection, invariant to rapid switch from darkness to brightness, consistent performance in low or even middle noise input situation.