Skip to main content
Product Roadmap • Phase 2 Development

Computer Vision Vehicle Detection

Scheduled for Phase 2 of the MoviSabio product roadmap, this module transforms existing municipal CCTV camera networks into real-time traffic density sensors using edge-deployed YOLO models—eliminating the need for expensive pavement cuts and inductive loop installation.

Inspect Vision Pipeline ↓ GDPR/LGPD Privacy Model Join Phase 2 Pilot Program →
📅

Roadmap Placement: Phase 2 Active Development

Targeted Edge Validation & Hardware Co-Design

While MoviSabio's core SUMO/TraCI micro-simulation and AITCS coordination algorithms are production-ready in simulation, the Computer Vision Vehicle Detection module is formally classified as Phase 2 in the product roadmap. Prototype models have been benchmarked on standard datasets; field edge integration is scheduled for municipal pilot hardware deployments.

Key Computer Vision Capabilities

Engineered to convert noisy optical CCTV feeds into structured, privacy-compliant traffic telemetry.

🎯

YOLO-Based Vehicle Detection

Utilizes lightweight YOLOv8 nano and small models quantised to INT8 precision for low-power edge compute (NVIDIA Jetson Nano/Orin or Intel OpenVINO edge gateways).

🚛

Multi-Class Classification & Counting

Distinguishes passenger cars, city transit buses, heavy commercial trucks, motorcycles, bicycles, and crossing pedestrians to calculate accurate Passenger Car Units (PCU).

🛣️

Lane-Wise Traffic Analysis

Software-defined polygonal detection zones isolate dedicated turning lanes, through-lanes, and bus rapid transit (BRT) corridors to compute specific lane queue depths.

⚡

Speed Estimation & Incident Alerts

Estimates approaching approach velocity vectors using calibrated ground-plane homography matrices, detecting sudden stalls, wrong-way movements, and stalled vehicles.

📡

CCTV & IP Camera Input Ingestion

Connects directly to existing municipal RTSP and ONVIF optical video streams. Operates at variable frame rates (10–30 FPS) to balance inference latency with network bandwidth.

🔄

Direct Integration with AITCS Core

Extracted vehicle counts and arrival times are immediately packaged into lightweight JSON payloads and transmitted via local MQTT to the signal coordination algorithm.

European GDPR & Brazilian LGPD Compliance

MoviSabio's computer vision architecture is strictly designed as an edge-only telemetry converter, not a surveillance system.

Immediate Pixel Discard

Raw video frames are held in volatile edge memory only for the duration of model inference (under 35ms). Once bounding boxes and class vectors are extracted, frames are immediately deleted from edge memory. No video recordings, facial images, or license plates are ever stored on disk or transmitted to the cloud.

Anonymized Vector Telemetry

The only data leaving the intersection cabinet is anonymous numeric telemetry: {"lane": 2, "count": 14, "class": "car", "avg_speed_kmh": 46.2}. This guarantees 100% compliance with strict international data protection laws, enabling seamless public-sector procurement.

Stakeholder Relevance

Discover how Phase 2 Computer Vision enhances each partner's operational horizon.

Public Sector

No Trenching, Zero Road Closures

Traditional loop detectors require road cutting and weeks of lane closures. Optical edge detection uses existing municipal camera poles with zero asphalt disruption.

Hardware Partners

Edge Device OEMs

We collaborate with edge compute vendors (NVIDIA Jetson, Hailo, Qualcomm) to package optimized containerized inference models for roadside ITS cabinets.

Investors

Roadmap Upsell Potential

Computer Vision represents a high-margin ARR expansion module for existing MoviSabio Smart City accounts, increasing annual contract value (ACV) by 35–45%.