Gemini AI

Inspiration

As a team of data scientists we wanted to tackle the biggest challenge going. That's the one involving LIDAR data.

What it does

Our project has three parts:

• A Neural Network to process and classify the LIDAR data into objects (e.g. foliage, tracks, buildings etc.)
• An interactive map showing the locations of the assets and the locations of the LIDAR scans
• A dashboard to bring this all together

How I built it

• Neural Network: Python + a pre-trained Pointnet Neural Network
• Interactive Map: Python & Folium
• Dashboard: Javascript, Bootstrap and React
• Other stuff: Coffee, caffiene & good vibes

Challenges I ran into

• The overhead LIDAR data wasn't of a high enough resolution so we instead used the train mounted LIDAR scan which performed better.
• We were unable to link the LIDAR and the Asset Register
• Anything to do with Devpost. Like we've got time to upload videos to YouTube!!!

Accomplishments that I'm proud of

• Just getting the Neural Network to classify things correctly was a huge challenge - we're very proud of this.
• Working well as a team and sticking to the requirement
• Our interaction with the mentors. They're all cool!

What I learned

• That AI in 48 hours is hard!

What's next for Gemini AI

• Sleep

Built With

• javascript
• python
• react