# Prince Tiwari

- **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:** [Prince Tiwari](https://cerebralvalley.ai/u/princetiwari19)
- **GitHub:** https://github.com/princetiwari1910/AuditLens
- **Demo video:** https://youtu.be/QPTpe_5HDx0
- **Gallery:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery
- **Page:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/2440

AuditLens: Making Safety Inspections Easier with AI

Walk through. Get proof. Fix what matters.

A blocked exit, a missing fire extinguisher, or a worker without proper safety equipment might seem like a small issue, but it can lead to a serious accident. Places like schools, hostels, clinics, restaurants, and workshops need regular safety checks, but frequent professional inspections are not always affordable or practical.

This is the problem we wanted to solve with AuditLens.

AuditLens turns a simple smartphone walkthrough video into a safety report that helps people understand what might be wrong in their surroundings and how they can fix it. Rather than giving users a generic AI response, it shows the actual video evidence behind each finding.

How AuditLens Works

1. Capture

Users upload a video recorded while walking through a facility. They can choose an existing safety checklist or add their own requirements in simple language.

2. Detect

Meta SAM 3.1 helps identify and segment important objects in the video, such as fire extinguishers, safety equipment, blocked exits, and other relevant elements.

3. Analyze

Meta Muse Spark examines the visual evidence and evaluates it against the selected safety rules. Each finding is marked as PASS, FAIL, or NEEDS REVIEW.

4. Show the Evidence

Every finding is linked to the relevant video timestamp and visual evidence. Users can see where an issue was identified and understand why it was flagged.

5. Recommend Fixes

AuditLens creates a list of practical corrective actions so users know what needs attention and what they should do next.

6. Track Improvements

Users can conduct another walkthrough and compare it with their previous audit to see whether the reported issues have been addressed.

What Makes AuditLens Different?

We did not want to build another application that simply detects objects or generates AI-written reports. We wanted to connect detection with actual safety requirements.

AuditLens brings together video processing, Meta SAM 3.1, Muse Spark, safety rules, and visual evidence in one workflow. The focus is on making every finding understandable and traceable rather than asking users to trust an AI-generated result without proof.

 Why It Matters

Many smaller organizations do not have the time or budget for frequent safety inspections. AuditLens aims to give them an easier way to identify potential problems and decide what needs attention.

Our goal is to help schools, hostels, clinics, workshops, and small businesses take safety more seriously without making the inspection process complicated.

We believe a safety problem should not just be reported. People should be able to see it, understand it, and know how to fix it.

AuditLens is designed as an AI-assisted pre-audit tool and supports, rather than replaces, professional safety inspections.

## More from Meta Global AI Developer Hackathon

- [StockSense Team](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/2437)
- [Avocaders](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/2438)
- [TBone](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/2439)
- [Screen-Time](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/2441)
- [S47](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/2442)
- [Latent](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/2443)

---

Markdown version of https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/2440. Site index for agents: https://cerebralvalley.ai/llms.txt · full text: https://cerebralvalley.ai/llms-full.txt