What is AI settop box Deep Runner?

AI-based general-purpose object recognition device 

that can easily implement AI systems without AI experts

Deep Runner AI set-top box is an edge-type AI device that provides the ability to detect or 

classify objects in real-time using artificial intelligence technology stand-alone without a server.

Supports easy and simple implementation 
of AI-based smart systems at low cost

Deep Runner is a small, palm-sized, low-power device that can be used as a key 

component for processing AI in various intelligent systems without an expensive AI system. 

In addition, general IT technicians can use it easily and conveniently without deep knowledge of AI.

Equips with the latest artificial intelligence algorithms.

Deep Runner Z1/Z4 models support the most complex and high-accuracy modern AI algorithms 

(MobileNet V2 1.4x / EfficientNet B2 / SSD / EfficientDet D0) and compute at high frame rates.

Supports high-precision computational resolution

Deep Runner does not transform the original algorithm and achieves high computational precision 

by operating in floating-point. Therefore, the high recognition rate of the original algorithm is achieved without deterioration.

※ The 16-bit floating-point computation result completely overlaps the 64-bit floating-point result (not visible in the graph).
※ The 16-bit floating-point computation result completely overlaps the 64-bit floating-point result (not visible in the graph).
Multi-model recognition

Deep Runner Z1 can connect multiple deep learning models to easily implement complex recognition scenarios.

< Example: Recognizing gender/age/facial expression from moving face >
< Example: Recognizing gender/age/facial expression from moving face >

<Example:  Recognizing gender/age/facial expression from moving face>

▶ Applications: AI signage system, Safety inspection, Vehicle identification, Product tag recognition, etc.

It executes high-resolution AI algorithms capable of recognizing fine objects

Deep Runner includes high-resolution pixel recognition algorithms 

so that it can simultaneously recognize very fine objects (up to 1080x1080 pixel images)

Simple Deep Runner Connection Configurations

The Deep Runner can be used by connecting to the system in a variety of ways.

1. Connect Deep Runner directly to a laptop computer and a USB camera

Directly connect the laptop to the Deep Runner via an Ethernet cable. 

The output screen can be monitored by accessing Deep Runner from the browser on your PC 

and the recognition results received in real time through Ethernet.

2. IP camera connection

HDMI signal can be used as input by using HDMI2USB conversion cable that comes with Deep Runner. 

 CCTV cameras can be connected using additional CCTV video equipment such as NVR.

3. Configuration using a router

Deep Runner connects to a router and thereby connects to external remote systems via the Internet.

The video output of Deep Runner can be transmitted through the web protocol 

so that the system can be monitored in real time even from multiple remote systems.

4. Minimal configuration

Camera is connected to the Deep Runner, and the recognition results controls an actuator or PLC device through the GPIO port.

During operation, users can connect a computer to monitor the system.

Specification table