Recognition of Train Driver's Attention Using Haar Cascade

Abstract

Driving a train is a very responsible task as it involves the safety and security of the train passengers. Though, railway department assures the presence of two drivers in the main cockpit of a train, but any human error could result in catastrophic fatalities. In order to ensure the attention of a rail driver, an indigenous solution is developed using a Raspberry Pi 3 B+, a webcam and Open CV libraries for the detection of driver's attentiveness. The algorithm works on the basics of Haar Cascade Classifier for capturing the eyes movement, which performs its operation using the binary classifier. Therefore, if the eyelids are close (i.e., the statement of the binary classifier is true) for more than 15 seconds, an alert will be sent to the railway control room regarding the driver's drowsiness, resulting the mitigation of disastrous outcome by ensuring the necessarily and timely measures. To validate the instrumentation, it was compared with Haar classifier along with a survey of the train drivers regarding the effectiveness of the developed prototype. From that survey, 70% of the drivers were satisfied with the effectiveness of the developed rail driver attention detection system.