SenseTime: Artificial Intelligence Extends Inward The industrial engine is starting to kick in

China Net Finance news on November 4, a few days ago, China's industrial powerhouse Qingdao held the 2021 World Industrial Internet Industry Conference. Participants generally believe that the development of Internet technology to the current stage, more in-depth penetration into the industrial manufacturing industry is the next step forward. During the conference, SenseTime, as a leader in the field of artificial intelligence, lost no time in launching the "SenseTime Industrial Engine", which provides a low-hanging fruit for how artificial intelligence empowers industrial manufacturing.

Massive computing power is the basis for coping with the "long tail" scene

Xu Li, co-founder and CEO of SenseTime, said in his speech, "At present, there are many 'long tail' scene needs in the industrial industry that have not been solved, such as the safety guarantee of workers, equipment, etc.; the management and maintenance of industrial assets, the control of product quality, and so on. In the process of digital and intelligent transformation, long-tail segmentation needs have great commercial value, and it is also the key to truly improving production efficiency and opening up the value closed loop. "There are many "long tail" scene needs in the industrial manufacturing industry that have not been solved, such as the safety guarantee of workers, plants, equipment, etc. in the factory, the management and operation of plants and equipment by energy companies, the control of product quality, etc. These application scenarios are not frequent, and there are many types. In the process of digital and intelligent transformation, enterprises usually give priority to and solve the head application with high concentration, and the long-tail application will be shelved because of factors such as cost, efficiency, and long-term development of the enterprise.

The new SenseTime industrial engine is to deal with a large number of "long tail" scenarios in industrial manufacturing as its own place. Relying on the core architecture of the new artificial intelligence infrastructure sensecore SenseTime, sensemars Mars Mixed Reality Platform and Sensefoundry enterprise SenseTime Ark Enterprise Open Platform, Industrial Engine is a comprehensive and intelligent industrial solution that can help industrial enterprises achieve the goal of cost reduction and efficiency increase by empowering various traditional industrial verticals.

The basis for responding to various long-tail scenarios with artificial intelligence is huge computing power support. SenseTime, which is under construction in Shanghai's Lingang District and will be put into use in early 2022, will become Asia's largest AI computing center with a terrifying computing power of 3,740 petaflops after completion. In figurative terms, it is possible to process a video that is 23,600 years old on Earth in 1 day. Based on this, the integration of AI chips and AI sensors can provide a powerful computing power foundation for SenseTime's engine, supporting the analysis, training and inference of massive data and algorithmic models.

Ai-power intervention has successfully increased efficiency for enterprises

Previously, SenseTime has accumulated a large number of solutions in the field of industrial manufacturing, which has become another basis for SenseTime to launch an industrial engine.

China FAW Group is a leading automobile manufacturer. In the metal stamping process, quality control inspectors have difficulty identifying small defects found in many different metal parts. Traditional inspection processes are time-consuming and lack consistency. SenseTime's original Enterprise Ark products enable customers to automatically detect more than 34 defects in 6 to 12 seconds, with a detection rate of more than 99%. Integrated into existing production lines, Ark can operate 24/7 with high stability, and the real-time detection function further helps customers identify defects or failures in assembly lines in real time, avoiding production process interruptions and potential losses caused by product defects.

Guangxi Pan Sugar Technology operates 10.6 million mu of sugar cane fields and 67 sugar mills. Traditional inspectors rely primarily on personal experience to detect impurities in sugarcane, such as sugarcane skins, leaves, and roots. To achieve automated inspections, SenseTime standardizes quality control processes by deploying Enterprise Ark. It can automatically detect various impurities, can achieve a detection rate of more than 90%, and greatly improve the detection efficiency by 67%.

High-speed rail is powered by overhead cables. The maintenance of power lines is critical to the day-to-day operation of high-speed rail and the safety of passengers. The Beijing-Shanghai high-speed railway operates and maintains more than 1,000 kilometers of railway network, with 130 million passenger traffic in 2020.

High-definition cameras are installed on customer patrol trains and regularly take photos of overhead cable catenarys as the vehicle travels along the railway. They generate millions of high-definition photos with location-encoded information for review analysis every quarter, requiring significant manpower to check for defects. SenseTime's Ark is deployed in the customer's data processing center. When hd photos are transmitted to the data center, the system automatically detects 2,160 defects in 514 components, including overhead cable equipment, support structures and suspension devices.

As of June 30, 2021, the Enterprise Ark platform has automatically detected a total of 26,914 defects. Compared to the traditional method of inspecting an average of 2.5 km of overhead cables a day, the new AI-driven inspection method can help customers inspect 50 km of overhead cables in the same amount of time. To improve operational efficiency, the customer has since expanded the scope of our AI inspection services, which cover more than 9,000 kilometers as of June 30, 2021.

Four-dimensional empowerment of industry to help the 14th Five-Year Plan goals

According to reports, the "SenseTime Industrial Engine" can empower the industrial manufacturing industry from four aspects:

A cost reduction and efficiency increase: to help enterprises from labor-intensive production mode to intensive and intelligent production mode, to solve the core problems of labor difficulties, high costs, low efficiency, etc., to achieve stable production capacity, cost reduction and efficiency increase, sustained growth;

Second quality control: according to the characteristics of products in different vertical fields, product defect detection can be customized, and the intelligent upgrading of industrial production lines can be realized, so as to ensure product production quality with higher accuracy and greatly improve the quality control rate;

Three safety management: intelligent safety management for workshops, plants, production lines, etc., early warning of dangerous areas and dangerous behaviors, thereby greatly reducing the accident rate of workers in the production process and comprehensively improving the safety factor;

Four intelligent O&M: Through ai, AR and other innovative technologies, we will carry out comprehensive and intelligent O&M of industrial production, build a digital twin of parks, plants, equipment, etc., realize the comprehensive digital management of physical assets, and reduce the threshold for learning and operation.

At present, the state in the "14th Five-Year Plan" clearly put forward to accelerate the process of industrial modernization, in 2025 manufacturing enterprises above designated size basically popularize digitalization, key industry backbone enterprises initially achieve intelligent transformation, by 2035, manufacturing enterprises above designated size will fully popularize digitalization, and backbone enterprises will basically achieve intelligent transformation.

SenseTime said sensetime's industrial engine has the ability to play a huge role in this process.

(Editor-in-charge: Zhao Rong)