Hotspotter

The landing page.
An example of an AI generated scenario image.
Users are then prompted to record themselves giving a scene size-up.
Feedback is provided based on realistic criteria and the user recording.

Inspiration

The scene size-up is one of the most important parts of the fireground response. Firefighters and officers on the first-due unit to a fire have the responsibility of providing following units with the crucial early information regarding the structure, the threats, and the tactics to suppress the fire. Firefighters aren't always able to get realistic practice giving scene size-ups, despite the important nature of this skill.

What it does

Hotspotter uses image generation AI to create a variety of structures and scenarios, so firefighters can get some exposure to giving a scene size-up for unfamiliar firegrounds. They are then able to record themselves giving a size-up, as if they were keying their mic to transmit it to other units. Hotspotter then uses generative AI to give adaptive feedback, based on whether any crucial pieces of information were omitted.

How we built it

Flask is a favorite of ours, being very lightweight and easy to work with when it comes to new frameworks or APIs. MongoDB serves our datalayer, and the frontend is vanilla HTML/CSS. Whisper a general-purpose speech recognition model from Cloudfare was used, to understand speech from recordings and translate to text.