Overview
Team
AI software engineer,
DevOps engineerSolution
IoT application
Country
United States
Industry
Public safety, surveillance monitoring
Services used
Discovery phase, AI software development
Time
March 2025 – today
Project idea
Municipalities invest heavily in street surveillance and public infrastructure, yet vandalism incidents like graffiti still slip through because monitoring is mostly reactive. CCTV footage is typically reviewed after a complaint is filed – when the damage is already done and the chances of identifying the person are low. Even when cameras capture the event, finding the right moment in hours of video is a slow manual work.
The core challenge is that motion detection isn’t specific enough. Streets are full of “normal” movement – pedestrians, cars, animals, changing light – so movement-based triggers generate noise and false alarms. At the same time, municipalities can’t justify 24/7 human monitoring. Besides, they often need to work within existing camera infrastructure and limited response capacity.
SecureWall AI was initiated to close this gap by adding an audio-based signal to the workflow. Shaking the graffiti stick has a distinctive sound profile, and recognizing it in real time makes it possible to detect graffiti right before it’s about to happen.
At the same time, the cameras are set to automatically capture the relevant video fragment, and instantly notify responsible teams via email/SMS. The result is a practical, scalable way to turn city cameras into an early-warning system – starting with a municipal deployment in one of the U.S. cities.
Challenges & Solutions
- Defined the real-world vandalism detection challenge with the client and clarified what “success” looks like for a municipal rollout.
- Completed the IoT application workflow mapping – from sound detection to evidence capture to notifications – so city teams can act quickly.
- Agreed on the essential capabilities for the first release and set clear priorities to keep delivery focused.
- Created a practical implementation plan that fits timeline and budget constraints while leaving room for future scaling.
SecureWall AI: key features of sound recognition software
- The software connects to street video monitoring cameras, and reads live streams when the particular sound is detected.
- It separates audio from video, and runs continuous audio analysis to recognize a spray paint can sound pattern.
- It triggers an event only when detection meets confidence/threshold rules to avoid noisy false alarms.
- Anti-vandalism software records video in rolling segments so the system can preserve the moment someone starts spray-painting.
- When an event is detected, it extracts the relevant fragment (with configurable pre/post context) and packages it as a clip.
- It attaches event metadata from the sound activated camera (timestamp, camera identifier, detection score) to support quick triage.
- Automated alerts via email and SMS with a direct link to the evidence clip and key context.
- Structured event history for auditing and performance review (detections, delivery status, and timestamps).
- Operational follow-up by making it easy to track repeated incidents by location and time patterns.
Integrations
