On February 23, MWC 2021, the annual global communications industry event, was held as scheduled at the Shanghai New International Expo Center. Hundreds of well-known manufacturers in the field of mobile communication showed the latest achievements in the field of mobile communication, such as 5G, artificial intelligence, Internet of Things and smart home. Inspur presented its 5G, smart computing and other solutions at the conference. At the meeting, Shoufang and Shanghai Gaoheavy Information Technology Co., LTD. (hereinafter referred to as: Gaoheavy Technology) jointly demonstrated intelligent protection solutions for production safety based on edge computing, providing 24-hour online "safety supervision" for high-risk industries such as construction and petrochemical.
The production of ten million, safety first! For construction, coal mining, metallurgy, petrochemical and other industries that need to carry out risky operations for a long time, safety is an eternal theme. Whether to wear safety protective equipment such as safety helmet and air breathing apparatus, whether there is open flame smoke in the working area, and whether there is personnel intrusion in the dangerous restricted area are important prevention contents of production supervision and safety management. However, the current industry field safety supervision work is still mainly manual management, completely rely on human eyes through the monitoring system or field inspection for verification, can not do 24 hours uninterrupted control, it is easy to appear omissions. Therefore, the use of computer vision, edge computing and other advanced technological means, intelligent supervision of the whole process of production has become the priority development direction of all industries to ensure safety.
In the exhibition area, the high-tech engineer introduced to the audience: "To achieve AI intelligent supervision based on computer vision, on-site video can be collected through the camera, relying on the safety hat monitoring, protective clothing monitoring, face detection algorithm to judge whether the behavior of workers meet the safety norms, and then automatically alarm and record. However, the challenges of this approach are clear: AI reasoning is computationally-demanding and often takes place in the cloud or in data centres. This is not acceptable for outdoor work scenarios, both in terms of cost and in terms of stability and availability of network environments."
Based on the industry-leading Inspur edge computing platform and advanced computer vision technology of Gaoheavy Technology, Inspur and Gaoheavy Technology jointly launched intelligent production safety protection solutions based on edge intelligent computing, and built a production safety control system. It is capable of monitoring and early warning of violations and dangerous events in outdoor work sites in construction, coal mining, metallurgy, petrochemical and other industries, showing great application value in the industry.
The solution is equipped with EIS200, Inspur's microserver, to build a high-performance edge reasoning computing platform, which can support the computing requirements of algorithms such as hard hat monitoring, protective clothing monitoring and behavior recognition. By extending computing power from the cloud to the edge, the solution not only helps enterprise users reduce investment in network and cloud computing infrastructure, but also reduces application latency and accelerates data analysis by processing visual data nearby.
In addition, Inspur EIS200 has strong adaptability to harsh outdoor environments, providing wide temperature support ranging from -45 ° C to 65 ° C and industrial-grade stable design, which can adapt to harsh outdoor environments and improve server stability and availability. EIS200 also reduces operation and maintenance stress through large-scale deployment and unified O&M management.
At the same time, this scheme integrates the profound accumulation of high-weight technology in the field of computer vision. The deep learning-based personnel wearing helmet recognition algorithm and human body detection algorithm have the advantages of high recognition rate, strong adaptability to the environment, head and shoulder detection model is not affected by human body shielding, not affected by human posture, good compatibility, etc. It can also identify other relevant dangerous behaviors based on dynamic video in real time, such as not wearing protective clothing, answering mobile phones, abnormal open flame, abnormal fall to the ground and so on.
This solution adopts cloud-edge collaborative architecture, which can carry out real-time, high-precision processing and query statistics of multi-objectives and multi-actions based on video structured analysis technology. It can continuously optimize the model and continuously improve the identification accuracy, which is an important protection scheme for production safety.
"At present, the solution has been successfully applied in a number of large enterprises in the construction, petrochemical, coal and other industries, helping customers greatly improve the safety protection capability of outdoor operations," said Chen Zhongping, General manager of Edge Computing Division of Gaozhong Technology. "In the future, We will also carry out deeper innovation and cooperation with Inspur in the fields of multi-channel face recognition, multi-mode UAV monitoring, airport runway panoramic stitching and other fields, and promote the landing of edge AI computing solutions in a wider range of application scenarios."
