Transforming traditional CCTV monitoring into intelligent, real-time security with AI-powered intrusion detection and automated video evidence management.
Industry: Automotive Manufacturing
Location: Jaipur, Rajasthan, India
Solution: Snoh Vision
Deployment: June 2026
Project at a Glance
| Project Detail | Information |
|---|---|
| Industry | Automotive Manufacturing |
| Application | Intelligent CCTV Surveillance |
| Cameras Monitored | 9 IP Cameras |
| AI Model | Custom-trained YOLOv11m |
| Detection Classes | Person and Guard |
| AI Model Performance | 0.784 mAP50 |
| Video Evidence | 15-second H.264 clips |
| Deployment | Cloud-hosted platform |
Overview: Making Industrial Surveillance Smarter with AI
Security monitoring is an essential part of manufacturing operations. From protecting equipment and inventory to monitoring restricted areas, manufacturing facilities need reliable surveillance systems that can identify potential security incidents without depending entirely on manual observation. However, traditional CCTV systems often provide limited intelligence. Security teams must continuously monitor multiple camera feeds, identify suspicious activity themselves, and manually retrieve footage whenever an incident occurs.
To address these challenges, SnohAI implemented an AI-powered video surveillance solution for an automotive manufacturing facility in Jaipur, Rajasthan. Powered by Snoh Vision, the customized system integrates nine IP cameras with a deep learning-based detection model to identify people, distinguish security guards, generate real-time alerts, and automatically capture video evidence.
The result is a centralized intelligent surveillance environment that helps security teams monitor facility activities, respond to potential incidents faster, and manage security evidence more efficiently.
The Challenge: Moving Beyond Manual CCTV Monitoring
Before implementing Snoh Vision, the facility faced several limitations with its existing surveillance approach.
1. Continuous Manual Monitoring
Security personnel depended on manually observing multiple CCTV feeds. Monitoring nine cameras simultaneously made it difficult to maintain consistent attention and identify every relevant incident.
2. Limited After-Hours Surveillance
Unauthorized presence in restricted areas during non-working hours could go unnoticed. The existing setup also lacked an intelligent way to distinguish security guards from other individuals during patrols.
3. No Automated Video Evidence Collection
Security incidents required manual footage retrieval and documentation. The absence of automatically generated, timestamped evidence made incident reviews and reporting more time-consuming.
4. Disconnected Camera Monitoring
The nine IP cameras operated without a unified monitoring interface, making it difficult to maintain a consolidated view of surveillance activities across indoor and outdoor areas.
5. Frequent False Alerts
Security patrols and static objects, including bags and equipment, could trigger unwanted notifications. Without intelligent filtering, repeated false alerts affected the efficiency of security monitoring.
6. Limited Access Management
The previous system lacked role-based permissions and a complete activity log, limiting visibility into who accessed, reviewed, or acknowledged security alerts. The facility needed a solution that could do more than record video. It needed intelligent monitoring, relevant notifications, automated evidence collection, and centralized control.
The Solution: Introducing Snoh Vision
SnohAI developed and deployed a customized surveillance solution using its enterprise computer vision platform, Snoh Vision.
The implementation combined AI-based person detection, intelligent alert processing, automated video recording, and a centralized cloud dashboard.
Instead of relying solely on security personnel to watch camera feeds, the system continuously analyzes camera footage and generates notifications when relevant detection events occur.
1. Custom AI Model for Intelligent Person Detection
At the core of the implementation was a custom-trained YOLOv11m deep learning model designed with two detection classes: Person and Guard. This enabled the system to differentiate between security personnel and other detected individuals, supporting more relevant security monitoring.
Key capabilities:
- Real-time person detection across all nine IP cameras.
- Guard-versus-person classification.
- GPU-accelerated inference using CUDA and FP16 precision.
- Simultaneous processing of all nine camera feeds.
The custom model achieved an mAP50 score of 0.784 during the reported evaluation.
2. Intelligent Alert Processing and False-Positive Filtering
Detecting activity is only one part of effective surveillance. A useful security system must also reduce unnecessary notifications. SnohAI implemented a multi-layer alert engine that applies several filtering mechanisms before generating alerts.
Key capabilities:
- Real-time detection notifications.
- Automatic snapshot capture for detected events.
- Confidence-threshold-based filtering.
- Aspect-ratio and area-ratio filtering.
- Guard suppression logic.
- Camera-specific working-hour schedules.
- Zone-based cooldown mechanisms.
- A 30-minute camera-level cooldown to prevent duplicate alerts for the same presence event.
Outdoor cameras were configured with camera-specific working-hour schedules, helping address unwanted night-patrol notifications.
3. Automated Video Evidence Management
The solution automatically captures visual evidence whenever a detection event occurs. Each event includes a reference snapshot and a 15-second H.264-encoded video clip, providing security teams with additional context for reviewing incidents.
Key capabilities:
- Timestamped snapshots.
- Automatic 15-second video clip generation.
- Browser-based playback with seek and replay functionality.
- Downloadable video evidence.
- HTTP Range streaming for smoother video playback.
This reduces the need to manually search through recordings whenever a potential incident needs to be investigated.
4. Centralized Cloud-Based Surveillance Dashboard
To simplify monitoring and system management, SnohAI deployed a centralized cloud-hosted dashboard connecting all nine surveillance cameras. The dashboard provides security personnel and administrators with a consolidated view of camera activity and detection alerts.
Key capabilities:
- Real-time WebSocket notifications without requiring page refreshes.
- Centralized monitoring of all nine cameras.
- Role-based access control with Admin and Viewer permissions.
- Timestamped user activity logs, including IP addresses.
- Alert filtering by camera, date, status, and severity.
- CSV export for structured alert reporting.
- HTTPS/SSL-secured deployment.
Surveillance Coverage: Nine Connected Camera Locations
Snoh Vision was deployed across nine surveillance nodes covering indoor storage facilities and outdoor operational areas.
| Camera Location | Coverage Area | Zone |
|---|---|---|
| MNT Store | Maintenance parts storage | Indoor |
| MNT/PPC Store | PPC and production planning store | Indoor |
| Die MNT. Store | Die maintenance storage | Indoor |
| Weld Store 1 | Welding consumables and assembly | Indoor |
| Weld Store 2 | Welding production area | Indoor |
| Out Side Die MNT. | External die maintenance zone | Outdoor |
| Road Side Die MNT. | Road-facing perimeter | Outdoor |
| Back Side of Warehouse | Warehouse rear access | Outdoor |
| Outside Scrap Yard | Scrap storage and disposal area | Outdoor |
By connecting these camera locations through a single platform, the implementation established a more organized and centralized surveillance environment.
Technology and System Architecture
The implementation combines computer vision, deep learning, backend processing, database management, video streaming, and secure infrastructure.
| Technology Layer | Technologies Used |
|---|---|
| AI / ML Engine | YOLOv11m, PyTorch, CUDA, FP16, Ultralytics |
| Backend | Python 3.11, FastAPI, SQLAlchemy, WebSocket, REST API |
| Database | MySQL 8.0 |
| Frontend | React 18, Vite, Tailwind CSS |
| Video Processing | OpenCV, H.264/avc1, HTTP Range Streaming |
| Infrastructure | Docker, Nginx Reverse Proxy, HTTPS/SSL, Linux Server |
| Security | JWT Authentication, Secure Cookies, RBAC, bcrypt, Audit Logs |
Results and Business Impact
The implementation introduced intelligent surveillance capabilities across the facility, replacing several manual monitoring activities with automated detection, notifications, and evidence management.
Before and After Snoh Vision
| Operational Area | Before Implementation | After Implementation |
|---|---|---|
| Intrusion Detection | Manual monitoring | Automated AI detection |
| Alert Response | Hours, if noticed | Real-time alerts within seconds |
| Video Evidence | No automated evidence collection | Automated snapshots and 15-second video clips |
| False Alert Filtering | No intelligent filtering | Multi-layer AI filtering |
| Camera Monitoring | Manual per-camera monitoring | Centralized dashboard |
| User Accountability | No complete audit trail | Timestamped activity logs |
| Access Management | Unrestricted access | Admin and Viewer roles |
| Incident Reporting | Manual and delayed | Instant alerts and CSV export |
| Night Patrol Alerts | Frequent false alerts | Reported as eliminated through scheduling and suppression |
Key Business Outcomes
Improved Security Visibility – AI-powered monitoring helps identify relevant activity across all nine cameras, providing security personnel with better visibility into facility operations.
Faster Incident Awareness – Real-time notifications help security teams become aware of detection events within seconds rather than relying entirely on manual camera observation.
More Efficient Alert Management – Multi-layer filtering, guard suppression, and camera-specific schedules help reduce unnecessary notifications and improve the relevance of security alerts.
Centralized Surveillance Operations – A single cloud-hosted dashboard brings camera monitoring, alerts, user permissions, and reporting into one integrated interface.
Automated Evidence Collection – Timestamped snapshots and downloadable video clips make it easier to review incidents and maintain supporting documentation.
Improved Access Control and Accountability – Role-based permissions and activity logs provide better control over system access and improve visibility into user actions.
Client Feedback
“The AI CCTV surveillance system deployed by SnohAI has fundamentally transformed how we manage security across our manufacturing facility. Real-time intrusion alerts with automatic video evidence have significantly improved our security response capability. The centralized dashboard gives us complete visibility across all cameras, and the intelligent false alert filtering means our security team can focus on real threats rather than chasing false alarms.” – Management Team, Automotive Manufacturing Client
Conclusion: Building Smarter Industrial Security with Snoh Vision
This implementation demonstrates how AI-powered computer vision can enhance traditional CCTV infrastructure and support more intelligent security operations in manufacturing environments. By combining custom-trained object detection, real-time alerts, intelligent filtering, automated video evidence, and centralized monitoring, Snoh Vision helped transform a manually monitored camera network into an integrated AI-powered surveillance system.
For manufacturing facilities looking to improve security visibility, streamline incident reviews, and make better use of their existing CCTV infrastructure, Snoh Vision offers an intelligent approach to modern surveillance.
About SnohAI
SnohAI is the AI products brand of Snohbricks Technology Pvt. Ltd., based in Jaipur, Rajasthan. It develops enterprise-focused AI solutions that help organizations improve operations, enhance security, automate processes, and make more informed decisions.
Its product portfolio includes:
- Snoh Vision: AI-powered computer vision and intelligent video surveillance.
- Snoh Fusion: Intelligent document processing and automated data extraction.
- Snoh Ava: AI-powered business intelligence and conversational data interaction.
- Snoh Docs: Document management and centralized information handling.
- Snoh Flow: Workflow and business process automation.
Explore Snoh Vision: AI Video Analytics Platform for Existing CCTV | SnohAI
Looking to make your existing CCTV infrastructure smarter?
Connect with SnohAI to explore AI-powered surveillance solutions for your business.