-
Pipes of different materials
Inspiration
We wanted to improve work for the hard working field technicians. In particular, we talked with the NWG members at the event to understand their challenges, noting quick material detection and a simplified app for operators was important.
What it does
The app predicts pipe material based on image and other file keepings, enabling a one-scan detection. It also allows operators to view and quickly update information about the assets owned by NWG and their location.
How we built it
We developed an app using Android Studio, working with Java as the background language. The machine learning models were created in Python.
Challenges we ran into
Our dataset was limited and hard to connect. We had troubled using a webserver to connect parts of our app together. Training deep ML models took a long time, especially for large sets.
Accomplishments that we're proud of
Our model accurately predicts Metal and Concrete pipe materials with 82% accuracy.
What we learned
Learnt to use Android Studio and work with Java, plus image processing with the relevant libraries.
What's next for NWG Operator App
We would further develop our crack detection capabilities, and make our app overall more clear.
Log in or sign up for Devpost to join the conversation.