# FoxWorld

- **Event:** [Meta Global AI Developer Hackathon](https://cerebralvalley.ai/e/meta-ai-developer-hackathon)
- **When:** Oct 26, 12:00 PM – Nov 4, 9:00 PM (EST)
- **Where:** Online
- **Team:** [Andres Valencia](https://cerebralvalley.ai/u/TWFoxcrack)
- **GitHub:** https://github.com/Foxcrack/viaguard
- **Demo video:** https://www.youtube.com/watch?v=pgNsmuAlP-E
- **Gallery:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery
- **Page:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/1397

Viaguard turns phone photos of streets into geolocated, severity-scored inspection reports for city teams. In Cali, Colombia, potholes can stay unrepaired for months because no city can inspect every street, and turning a photo into a located, prioritized work order is slow and manual.

Muse Spark reads the scene and suggests what to look for. SAM 3.1 segments each defect, and also finds pedestrians and license plates to blur them before anything is stored (privacy blur is on by default). Severity and urgency are computed by deterministic code from the measured area, not invented by the language model, and Muse Spark writes the repair recommendation using only what was measured. When the measured severity and the visual condition disagree, Viaguard shows a note instead of hiding it. Each inspection becomes a pin on a civic map and a printable report.

Tested on 10 photos I took in Cali: SAM detected at least one defect in 7 and none in 3. It is conservative and can miss diffuse wear, and areas are relative to the image frame, not square meters. A secondary construction before/after module is included but not validated on real field photos.

Built solo during the hackathon with an AI coding agent that I directed and tested.

## More from Meta Global AI Developer Hackathon

- [Slopbook Relay](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/1394)
- [MEDSIM](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/1395)
- [Shankhya AI](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/1396)
- [Ajd](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/1398)
- [mtkcodex](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/1399)
- [RealityReplay Labs](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/1400)

---

Markdown version of https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/1397. Site index for agents: https://cerebralvalley.ai/llms.txt · full text: https://cerebralvalley.ai/llms-full.txt