# Gateward

- **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:** [Chetan Tai](https://cerebralvalley.ai/u/chetan_tai)
- **GitHub:** https://github.com/Chetan-Tai/gateward
- **Demo video:** https://youtu.be/t1sOOCgJw4M
- **Gallery:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery
- **Page:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/1617

Gateward is a spatial accessibility auditor that turns a handful of smartphone photos of a building entrance into a measured, ADA-referenced accessibility assessment.

Accessibility audits of entrances, checking door clear width against the ADA’s 32-inch minimum, spotting steps, ramps, obstructions, and hardware that requires grasping or twisting, are normally done by hand, with a tape measure and a trained eye. Gateward automates that process without sacrificing the measured rigor: it doesn’t just describe a photo, it calculates a real physical measurement from it.

The pipeline uses SAM 3.1 to segment the door and a standard sheet of printer paper taped to it as a known-size calibration reference, converting pixel-space detections into a real-world measurement in inches with a stated confidence range and a PASS/BORDERLINE/BLOCKING verdict. Muse Spark then reasons over SAM’s actual detection geometry, real bounding-box coordinates for the door, stairs, ramps, handrails, and other features across all submitted photos, to produce a structured set of accessibility flags with severity levels and remediation suggestions.

The result: a property manager or business owner can self-check an entrance in minutes without hiring an auditor, and a professional auditor can use it to screen many entrances quickly, reserving manual review for the cases that come back borderline.

## More from Meta Global AI Developer Hackathon

- [KisanAi](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/1614)
- [Gifty Tech](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/1615)
- [Nikita goyal](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/1616)
- [StackOne](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/1618)
- [KaiSai](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/1619)
- [Apprentice Coach](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/1620)

---

Markdown version of https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/1617. Site index for agents: https://cerebralvalley.ai/llms.txt · full text: https://cerebralvalley.ai/llms-full.txt