Road crashes are the single greatest cause of death of healthy Americans and most accidents occur just miles away from one’s home. We decided we wanted to inform the average driver of possibly dangerous intersections and roads based on different outside circumstances. Drivers could then make cognitive and smart decision, avoiding hazardous areas and driving to their desired destination safely.
What it does
Our website provides drivers with an easy to use, interactive map, depicting all possible roads considered hazardous in given area. Based on previous crash data, our website provides predictions of what percent of the time a certain road will be considered "dangerous". Since all the information on our website is neatly presented and easy to understand, users are able to make more informed decisions about what route is safest to travel given certain conditions.
How we built it
Our website uses Microsoft Cognitive Services to identify problematic intersections and pieces of road. We used a database of car accidents across the US and found satellite images of the locations of these crashes. We then used the images of intersections with previous history of crashes to train a model to recognize potentially dangerous streets and intersects based on outside factors.
Challenges we ran into
Figuring out how to implement the Google Maps API into our HTML website was one of the challenges we faced, along with finding a way to customize the map to add our own pin points to certain locations.
Accomplishments that we're proud of
Pulling together all of the backend information with the front end design helps give the user a nice visual and also provides accurate information so they stay safe and make informed decisions on possible courses of travel.
What we learned
Microsoft Cognitive Services was new to us before the start of this project. We learned how to train models to recognize hazards based on the data we collected.
What's next for Collision Vision
In the future we plan to incorporate weather predictions and how that might effect road conditions in certain areas. We also plan on making the information even more accessible by expanding to a mobile app so drivers can view dangerous road predictions during their travels.