Inspiration
Our startup Aurum AI is building intelligence and security monitoring software to enable "eyes-off" automation of scanning for security threats. This is going to be an early prototype.
What it does
The system can connect to different video feeds (currently only pre-recorded) and take in free-form categories to be alerted of (tank, smoke, soldier, truck, etc.) and then give you alerts when the category is found in the video.
How we built it
Next.JS / Mapbox for the front end, Radiant / GPT turbo + vision, python for the backend
Challenges we ran into
Integration :) multiprocessing in python, latency of cloud-based models
Accomplishments that we're proud of
Getting a prototype in 24 hours!
What we learned
How good LLM based analysis of video is and how precise it can be
What's next for Aurum Global Threat Analysis System
Continue building our capabilities into an MVP for customers
Log in or sign up for Devpost to join the conversation.