As a native Texan, I empathized deeply with families who were impacted by WWD fatalities in my area, and I felt compelled to find ways to prevent future tragedies. Through extensive research, I discovered that while states and government agencies have made significant investments in developing WWD warning systems along roads, fatalities from these types of accidents have not decreased.
To tackle the issue of wrong-way driving, I have developed an advanced in-vehicle dashcam system powered by a Raspberry Pi. The system leverages a front-facing camera to capture live video feed, which is analyzed in real time using a custom-trained EfficientDet object detection model. This model is specifically fine-tuned to detect the backs of road signs, which indicates potential wrong-way driving behavior.
By monitoring the frequency and intervals of detections, the system determines whether the vehicle is traveling in the wrong direction. Upon identifying a potential wrong-way driver, it immediately sounds a continuous alarm aimed at capturing the driver’s attention and prompting them to stop the vehicle.