YM Edge Box

   YM020-SC-B050 is equipped with domestic deep learning chip BM1684 and an 8-core processor A53@2.3GHz. With deep learning computing power of up to 17.6Tops, it can process 20 channels of HD video at the same time and support 2-channel encoding. The overall power consumption is less than 25W. Thirteen general algorithm scenarios are built-in, including face recognition, safety helmet identification, safety belt identification, safety vest identification (key personnel), clothing detection (short sleeves and shorts), smoking detection, open smoke detection, open fire detection, electronic fence intrusion (illegal intrusion), electronic fence crossing (illegal crossing), overload detection, vacancy detection and personnel counting. It accurately covers the application scenarios of intelligent energy, construction site and park.


Specification and Characteristic

Product Model
Application Scenarios

Intelligent construction site, park, factory, chemical industry, energy, community, security, business hall, scenic spot and  venue

Deep Learning Processor
Deep Learning Performance
17.6T@INT8 / 2.2TFLOPS@FP32
Master Processor
8-core ARM A53, basic frequency up to 2.3GHz
Video Decoding Capability
Support 20 1080P video decoding
Video Encoding Capability
Support 20 1080P video encoding
Support SATA storage, support 2TB storage
Display Interface
1 * HDMI
Other Interface
1 * RS-232 / 1 * RS-485/Self-defined I/O
180mm X 180mm X 60mm
Other Parameters

1 * HDMI, SATA, 1 * RS-232, 1 * RS-485, I/O interfaces

Partner Introduction

PHILOSIGHT Official Website

   Established in Beijing Zhongguancun Park, PHILOSIGHT is a technology-based company committed to creating a better future with science and technology.

   The team members are composed of computer vision experts from Columbia University, Johns Hopkins University, University of Birmingham and other international famous universities. They have won awards in international computer vision and machine learning-related contests several times.

   Since the establishment of the team, it has always adhered to the development mission of enabling safe production with deep learning vision technology and combined with computer vision algorithm, high-precision image measurement algorithm and other technologies. It has had an in-depth understanding of risk operations in power construction. At present, it has realized violation identification and early warning for various risk operation scenarios in the process of power infrastructure construction, which is widely used in provincial power transformation and line construction sites.

   As an innovator of deep learning vision technology, the deep learning intelligent video security early warning system we developed provides technical support for state-owned enterprises and large central enterprises such as State Grid, China National Nuclear Corporation, China South Industries Group Corporation, PetroChina, China Mobile, etc., and empowers deep learning in enabling safe production, solve existing and potential security risks, so as to reduce labor costs and protect safe production.

   The company takes computer vision and machine in-depth learning as the core technology. Based on face recognition, target detection, behavior dynamic recognition, target tracking and other technologies, the company lands various application scenarios to empower the industry. It has the ability of independent research and development of software and hardware and key technologies such as predictive analysis based on small samples, industry fault recognition based on visual perception, and scene violation recognition. It provides integrated system applications with industry data+ machine learning+ Deep Learning embedded software and hardware solutions as the core, and is committed to working with the energy industry to build a safe and bright future with Deep Learning.