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