ClearDrive

• 

  interface showing different routes

• 

Inspiration

Inspired by the alarming statistics of road accidents attributed to poor air visibility. In cities like Delhi and Mumbai, where low visibility contributes to a significant number of accidents annually, our team recognized the urgent need for proactive measures.

What it does

ClearDrive uses AVI or the Air Visibility Index which is a relative number formulated by machine learning algorithms to determine which road has a higher visibility in current situation. First we collect data from different image sources like cctv on roads, traffic signals and toll booths. Then after processing those images we obtain AVI and then interface it with opensource maps and produce the safest road to traverse, thereby averting to the danger of driving in poorly visible areas.

How we built it

There are three stages of building our idea:

1. We process different images taken from different coordinates and then find the AVI using opencv algorithms and machine learning model like yolov5. The calculated AVI is then sent to a mongo database which stores the AVI along with coordinates.
2. A leaflet js frontend is used for the map interface which returns list of coordinates between two given coordinates ( start and end point) to the backend.
3. A flask backend handles those requests and then using nearest neighbour search we find the AVI of the coordinates which are the nearest to the list of coordinates of roads between the end points we obtained from frontend. After that the AVI for the entire road is interpolated and the safest road is produced in the map.

Challenges we ran into

We ran into quite a lot of challenges like interfacing the AVI data in leaflet and finding those AVI which are in the range of coordinates produced by the frontend. We also faced challenges in finding AVI using ML models.

Accomplishments that we're proud of

We are proud of the solution and the cause behind which we are working.

What we learned

We learned a lot about Pytorch , opencv and leaflet.js.

What's next for ClearDrive

We will use better ML techniques to find AVI and use better routing techniques to get more paths with AVIs so that there will be more options for the drivers. Above all we will be needing more images and especially in real time like maybe getting images from dashboard cams and more cctvs so that we get accurate and real-time data of AVI.

Built With

• flask
• javascript
• leaflet.js
• opencv
• python
• pytorch
• yolov5