1. Purpose and Scope

This document sets out the operating environment and system requirements — covering server and client hardware, network, database, and supported cameras — for deploying and operating "MirAI-V2" (version 2.0.3), the on-premise AI Video Management System (VMS) provided by MarkAny Co., Ltd.

MirAI-V2 ingests video from IP cameras (ONVIF / RTSP), performs GPU-based AI inference (object detection, fire/smoke, fall-down, intrusion, and similar events), and provides recording, real-time notification, and monitoring through a dedicated client. An NVIDIA GPU is required for AI inference.

2. Server Requirements

The server is the core machine that runs the AI inference engine, video ingestion (RTSP capture), recording, the API / WebSocket / TCP services, and the database. GPU performance directly determines the number of cameras that can be processed concurrently.

2.1 Minimum Configuration

2.2 Recommended Configuration

2.3 Required Software and Dependencies (GPU / Runtime)

Of the items below, the NVIDIA driver and CUDA Toolkit must be installed in advance, before running the installer. All other dependencies (PostgreSQL, VC++ Runtime, GStreamer, CUDA libraries, etc.) are installed automatically by the MirAI-V2 installer.

3. Client Requirements

The client is the dedicated application used for live grid viewing, AI pipeline editing, and reviewing notifications and incident logs. It can run on the same machine as the server (client + server setup) or connect from a separate machine.

4. Network & Port Requirements

MirAI-V2 uses the following ports for client/server communication and for communication with cameras. When the client and server are on different machines, allow inbound traffic for the relevant ports in the server-side firewall.

4.1 Service Ports (Client ↔ Server)

4.2 Server ↔ Cameras / Database

4.3 Connection Limits (Defaults)

All of the above are configurable and will be optimized during sizing according to the operating scale.

5. Database Requirements

MirAI-V2 uses PostgreSQL to store persistent data (users, camera settings, AI pipelines, incident logs, notification rules, audit trail, sessions, system settings, etc.). PostgreSQL is installed automatically by the installer, so no advance installation is required by the customer.

6. Supported Cameras

MirAI-V2 supports IP cameras that comply with standard protocols. No proprietary cameras are required, so existing surveillance camera assets can be reused.

7. Storage Sizing Guidance

The storage capacity required for recordings depends on the number of cameras, resolution, frame rate, codec (bitrate), recording mode (continuous vs. event-based), and retention period. For continuous recording, capacity is generally estimated by the relationship below.

8. Camera-Count Guidance

The number of cameras a single server can accommodate is determined primarily by GPU VRAM capacity and AI inference load (analyzed FPS). Capacity can be expanded with multi-GPU configurations by adding GPUs. The table below provides guidance by GPU memory size.

9. Browser Requirements (Viewing the Manual)

To view this document and the HTML-format MirAI-V2 manual, the latest versions of the following browsers are recommended. These are requirements for viewing the documentation, not operating requirements for the system itself.

10. Notes and Assumptions

• This is an on-premise system (installed on the customer's premises). AI inference is completed within the server; video data is not sent to any external cloud.
• An NVIDIA GPU is required for AI inference. AI features will not operate on non-NVIDIA GPUs or on systems without a GPU.
• The confirmed values in this document (OS, ports, GPU / CUDA, TensorRT, etc.) are based on the standard configuration of MirAI-V2 v2.0.3.
• Items shown as [Recommended: …] / [Guideline: …] placeholders are finalized during sizing based on the customer's operating conditions.
• The contents of this document are subject to change without notice due to product improvements and other factors.