VACS — Vehicle Access Control System

Project Overview

Developed for Universidad Tecnológica de Bolívar (UTB), VACS solves a specific institutional constraint: an open campus that cannot install turnstiles or physical barriers still needs to control what enters its secure zones. Rather than gating people, VACS focuses on vehicles — a camera at each access point uses YOLOv11 and OpenCV for automatic license plate recognition, streaming detections over RTSP into a lightweight FastAPI buffer on the camera device, which forwards events to a central NestJS API authenticated via an API key. The backend persists events in PostgreSQL, queues background jobs on Redis via Bull, and pushes live updates over Socket.IO to a Next.js dashboard where staff manage employees, vehicles, roles and permissions, incidents, and reports. Employee authentication runs entirely through Microsoft Entra ID SSO, with no local passwords. The infrastructure is segmented into an edge network (dashboard and reverse proxy) and a data network (database and queue), with only the backend allowed to cross between them, all behind a single nginx reverse proxy with TLS. The team also ran its own internal security audit, tracking findings such as a leaked Azure secret in an environment file and a missing permission guard on employee creation. The license-plate detection component is still a work in progress, and the default timezone is set to America/Bogota, matching UTB's campus in Cartagena, Colombia.

Scope & Capabilities

1. 01

   Automatic license plate recognition

   YOLOv11 + OpenCV computer vision reads plates in real time at vehicle entry points

2. 02

   Centralized access event logging

   every vehicle entry and exit is recorded as a single source of truth

3. 03

   Employee, vehicle & permissions management

   role-based administration from a central web dashboard

4. 04

   Incident tracking & reporting

   logging and reporting of access-related incidents

5. 05

   Enterprise SSO authentication

   employee login via Microsoft Entra ID, with no local passwords

6. 06

   Edge camera event buffering

   a lightweight local service on the camera device queues and forwards detections

7. 07

   Network-segmented architecture

   separate edge and data network zones, with the backend as the only bridge between them

8. 08

   Internal security auditing

   an in-house security review tracking and remediating findings such as leaked secrets and missing permission guards

Technology Stack