AI Edge Computing Server | Spark Taiwan｜Argo AI VMS & Edge AI Video Surveillance Solutions

AI Edge Computing Server

An AI-powered video management platform with distributed architecture and HTTPS encryption, offering high scalability, flexibility, and secure data sovereignty for diverse surveillance applications.

檔案下載

Download

產品介紹

Overview

VLM-Based Deep Recognition

Core On-Premise Computing, Unlocking Unlimited Value

Powered by Argo AI at the edge, the system leverages a Vision-Language Model (VLM) to deliver advanced visual intelligence.

Spark AI edge server enables on-premise deployment, transforming live video into structured, real-time actionable data. It detects objects, analyzes attributes, and understands contextual interactions for fast search, precise insights, and confident decision-making.

All video data is processed locally and remains within your network, ensuring strict data sovereignty and security for government and high-security industrial environments.

Core Capabilities

Automatically understands, analyzes, and interprets visual data—transforming video into actionable insights and decision-ready intelligence.

Deep Recognition

Triggered by Argo AI at the edge, leveraging a Vision-Language Model (VLM) for deep visual understanding.

Visual Search

Skip timeline search, one snapshot finds target attributes (e.g., clothing, vehicle type) instantly.

Data Indexing

Supports cross-system synchronization of recognition results.

Smart Alerts

Real-time notifications via LINE and email instantly alert you to critical events.

High-performance NPU

mbedded Kneron NPU with up to 400 TOPS for real-time on-premise video analytics, reducing energy use by 30% while ensuring privacy and data security.

AI Data Indexing

Syncs multi-host AI data, converting cross-camera and cross-site metadata into structured indexes for accurate correlation.

Deep Recognition

Accurately detects attributes, states, and interactions—such as identifying employees in yellow uniforms and flagging anomalies like moved fire extinguishers—for fast localization and smart decision-making.

Visual Search and feature extraction

Supports image-to-image search and feature extraction (e.g., clothing color, PPE equipment, and other special objects), quickly locating targets or anomalies from massive datasets for simplified analysis.

Real-time Notification

Supports real-time notifications via email and communication apps (e.g. LINE), enabling managers to stay updated on events and take quick action to mitigate

Applications

The evolution of AI technology is fundamentally reshaping the value of video surveillance.

Spark focuses on seamlessly integrating hardware architecture with AI algorithms, enabling deep visual understanding. We help customers accelerate innovation and transform systems from passive recording into proactive, real-time response.

Smart City / Traffic

Identify vehicle attributes &
summarize traffic events.

Construtin / Industrial

PPE safety & compliance auditing.

Public Safety

Personnel and Object Recognition
Enhances access control and anomaly detection.

Healticare

Medical PPE Compliance Detection
Ensures operational safety and regulatory compliance.

Product Specifications

Categories	Item	Content
AI Management	Applicable	32–128 channels
	Person Appearance	- Tops, Pants, Headwear, Eyewear, Carrying - Supports 10+ color detection
	Designated Object	- Industrial and Medical PPE - Special vehicles - Vessels
	Advanced Analytics	Fire and smoke detection
	Video Understanding	Automatic extraction of scenarios and behaviors
Server	Model	SR-VLM-330-V2 (KNEO330 plus)
	AI Accelerator	Kneron™ embedded NPU
	Computing	Power Total 400 TOPS
	Memory	32 GB DDR-4 3200 MHz
	Storage	2TB NVMe SSD
	Operating System	Ubuntu Linux
	I/O port	Front: USB 2.0 x2 , Rear: USB 3.2 Type-A x4/Type-C x1
	Network	RJ-45 Gigabit Ethernet x1
	Accessories	Rack Mount Ears
	Power Supply	100-240V 50/60Hz, Max 850W
	Country of origin	Made in Taiwan
	Dimensions	428 x 350 x 89mm (Rack mount 2U)
	Certification	NDAA / CE / FCC / BSMI
	Operating Temp	Operating Temp 0°C ~ 40°C

*This host must be used with Argo AI VMS platform to achieve full AI performance.

VLM-Based Deep Recognition

Core On-Premise Computing, Unlocking Unlimited Value

Powered by Argo AI at the edge, the system leverages a Vision-Language Model (VLM) to deliver advanced visual intelligence.

All video data is processed locally and remains within your network, ensuring strict data sovereignty and security for government and high-security industrial environments.

Core Capabilities

Automatically understands, analyzes, and interprets visual data—transforming video into actionable insights and decision-ready intelligence.

Deep Recognition

Triggered by Argo AI at the edge, leveraging a Vision-Language Model (VLM) for deep visual understanding.

Visual Search

Skip timeline search, one snapshot finds target attributes (e.g., clothing, vehicle type) instantly.

Data Indexing

Supports cross-system synchronization of recognition results.

Smart Alerts

Real-time notifications via LINE and email instantly alert you to critical events.

High-performance NPU

mbedded Kneron NPU with up to 400 TOPS for real-time on-premise video analytics, reducing energy use by 30% while ensuring privacy and data security.

AI Data Indexing

Syncs multi-host AI data, converting cross-camera and cross-site metadata into structured indexes for accurate correlation.

Deep Recognition

Accurately detects attributes, states, and interactions such as identifying employees in yellow uniforms and flagging anomalies like moved fire extinguishers—for fast localization and smart decision-making.

Visual Search and feature extraction

Real-time Notification

Supports real-time notifications via email and communication apps (e.g. LINE), enabling managers to stay updated on events and take quick action to mitigate

Applications

The evolution of AI technology is fundamentally reshaping the value of video surveillance.

Smart City / Traffic

Identify vehicle attributes &
summarize traffic events.

Construtin / Industrial

PPE safety & compliance auditing.

Public Safety

Personnel and Object Recognition
Enhances access control and anomaly detection.

Healticare

Medical PPE Compliance Detection
Ensures operational safety and regulatory compliance.

Product Specifications

Categories	Item	Content
AI Management	Applicable	32–128 channels
	Person Appearance	- Tops, Pants, Headwear, Eyewear, Carrying - Supports 10+ color detection
	Designated Object	- Industrial and Medical PPE - Special vehicles - Vessels
	Advanced Analytics	Fire and smoke detection
	Video Understanding	Automatic extraction of scenarios and behaviors
Server	Model	SR-VLM-330-V2 (KNEO330 plus)
	AI Accelerator	Kneron™ embedded NPU
	Computing	Power Total 400 TOPS
	Memory	32 GB DDR-4 3200 MHz
	Storage	2TB NVMe SSD
	Operating System	Ubuntu Linux
	I/O port	Front: USB 2.0 x2 , Rear: USB 3.2 Type-A x4/Type-C x1
	Network	RJ-45 Gigabit Ethernet x1
	Accessories	Rack Mount Ears
	Power Supply	100-240V 50/60Hz, Max 850W
	Country of origin	Made in Taiwan
	Dimensions	428 x 350 x 89mm (Rack mount 2U)
	Certification	NDAA / CE / FCC / BSMI
	Operating Temp	Operating Temp 0°C ~ 40°C

*This host must be used with Argo AI VMS platform to achieve full AI performance.

Back

Products

產品介紹 Overview

VLM-Based Deep Recognition

Core On-Premise Computing, Unlocking Unlimited Value

Core Capabilities

Applications

VLM-Based Deep Recognition

Core On-Premise Computing, Unlocking Unlimited Value

Core Capabilities

Applications

產品介紹

Overview