The Project
We decided to create ____ to help prevent and reduce the number of accidents on the road for drivers, bikers, and pedestrians along with predicting accident severity in order to help accident management processes such as first responder time and resource management ultimately saving lives.
Accident Heat Map
Our heatmap is able to show which streets have higher chances of an accident happening allowing drivers to take caution around those areas or possibly avoid them on their trips. This would prevent countless deaths as drivers would now know where they should be careful and try to avoid making mistakes. Our heatmap can also go hand in hand with our accident severity predictor as if drivers can see hotspots on the map and plug in the location and time to our predictor so if an accident were to happen they could see how bad it was.
Accident Severity
The project is to be able to predict the severity of an accident by the location and the time. A user can input a location and time, and based on specific locations, weather, time, and other features we are able to predict how severe that accident would be.
Distracted Drivers
The model intakes a picture of a driver from inside the car. If they are classified as distracted we can separate them into 4 of the most common reasons for being distracted. Using their phone, radio, drinking/eating, or using their mirror. An alert will go off, similar to seat belt alarms, and will not stop unless the model detects that the driver is no longer distracted. About 3,000 people die from distracted driving every year, with our model we would be able to reduce the number of distracted drivers immensely and save countless lives.
Facts
Statistics on how dangerous driving was in America in 2019.
Million Crashes
Thousand Deaths
Million Injured
Billion Dollars spent
Features
Our website's features and how they work
Accident Heatmap
Our accident heatmap uses a neural network in order to find hotspots for road accidents for a city selected by the user.
Accident Severity
Predicting accident severity based on location and time using a random forest classifier
Based on data from cities we are able to use time, weather, infastructure, and many other features in order to pinpoint certain hotspots for areas by the street.
Because accident severity depends on climate and location these features were crucial in getting a near perfect accuracy. We ask the user to input the location and time, and our model is able to find the patterns.
Our CNN model is able to detect distracted drivers and what they are doing. It classifies drivers as either attentive, using phone, using radio, eating/drinking, or using mirror.
