Safe Drive

Our mission is to make every public ride safe!

Comment

Inspiration

There is currently no system available that detects any anomaly taking place in the public transport sector. However, if transport service providers have access to a mechanism to get real-time information/feed regarding the mishaps, they can improve their services accordingly, improving the overall customer experience.

What it does

The model detects violence in a public transport vehicle and sends a push notification to the Driver and Central HQ to take action.

How we built it

The mobile application was built using Flutter, and the Deep Learning model was implemented using Python and Tensorflow. The training was done using Habana Gaudi DL1 instance by running the Docker instance of the TensorFlow installer.

Challenges we ran into

We faced challenges integrating the model into the application workflow and exploring and utilizing the DL1 Instance since we hadn't experimented with HPC before.

Accomplishments that we're proud of

An end-to-end system that can identify violence and send push notifications to a central headquarter and the driver, which works in real-time.

What we learned

->Integration of Deep learning models within server-sided applications. ->Connecting services like AWS-S3, Docker Instances, EC2 together via CLI and Code. ->The methodology to be followed when creating a deep learning model to detect violence. ->Building an end-to-end prototype that can provide impactful results.

What's next for Safe Drive

-> Scaling the app to run for masses -> Data analytics with the information logs -> Adding other detection models like Small handheld objects to detect guns and other illegal items -> The data can be shared with law enforcement agencies and government to take actions -> Improving the accuracy of the model by taking manual snippets of the incidents which weren't reported by the app.

Built With

Share this project:

Updates