Introduction:
In recent years, the internal and external environment of the surveying and mapping industry has undergone great changes, and in the face of the wave of the era of digital and intelligent transformation of the whole society, the industry is facing great challenges in technological transformation and upgrading.
Intelligent surveying and mapping is characterized by cross-border integration, ubiquitous perception, intelligent autonomy and precise service.
Founded in 2016, Xi'an Ruisi Digital Intelligence Technology Co., Ltd. (hereinafter referred to as "Ruisi Digital Intelligence"), with artificial intelligence, big data, and cloud computing as the research direction, has long been committed to the intelligent interpretation and analysis of satellite remote sensing image data, based on the field of surveying, mapping and geographic information, to provide intelligent solutions for enterprises and institutions, to help them increase their value-added efficiency, improve decision-making level, and realize digital and intelligent transformation, and the service scope has gradually extended to natural resources, land surveying and mapping, agriculture, forestry and water conservancy, smart cities, digital villages, Energy, carbon peaking and carbon neutrality, electric power transportation, industrial manufacturing and other fields. This time, the reporter of "China Surveying and Mapping" interviewed Zhao Hang, President of RESSET, and asked him to talk about the intelligent surveying and mapping practice of RESSET in recent years.
Reporter: From the beginning of its establishmentCloud computingWhat are the opportunities for business to help the intelligent transformation of the industry at present?
Zhao Hang: At the beginning of its establishment, RESS was mainly engaged in cloud computing, which is divided into three layers: infrastructure layer, platform layer and application layer. When we first entered this field, many domestic government and enterprise customers had duplicate infrastructure construction and did not go to the cloud, and our goal was to help them build cloud computing centers, which was also a technology and application hotspot in the market at that time.
In this process, we put forward the "Four Syllogisms on Government and Enterprise Development" that are still applicable to this day: the first step is to migrate all services to the cloud and do digital transformation; The second step is the assetization of IT and data, and the creation of an assetization platform for data; The third step is to continue to do AI intelligence; The last step is to upgrade to a smart service center, which has the intelligent ability to provide internal and external decision-making through data analysis. This process is step-by-step, and cannot be carried out across levels.
Since 2019, the pan-natural resources, land surveying and mapping units with industry influence that we have come into contact with will go through such a process, and this process is basically closely integrated with the remote sensing business. Since the pure cloud computing power provided is not enough to improve their production efficiency, we gradually start to participate in the business level from the aspects of intelligent interpretation and recognition of remote sensing data, and improve their production efficiency from the business level by realizing the intelligent processing of data.
The fundamental logic of this transformation and development is to improve the production efficiency and effectiveness of government and enterprise customers as the core point. With this goal in mind, we provide corresponding technologies from the hardware foundation, storage and computing resources, and business level.
Reporter:Sharp and digitalWhat are the intelligent surveying and mapping services brought to the industry? Advantages in**?
Zhao Hang: At present, RESSET products cover IaaS, PaaS and artificial intelligence industry applications, based on "remote sensing AI + cloud computing + data storage" and other technical means, the self-developed RS Matrix matrix platform can realize the overall solution and application of software and hardware integration.
The platform includes five product modules: data storage and management platform, multi-computing power cloud platform, remote sensing AI intelligent interpretation platform, spatio-temporal data service platform, and cloud native application management platform.
The platform can realize modular and matrix product portfolio and service capabilities, and then form a series of application solutions such as intelligent surveying and mapping production, spatio-temporal data management, real-world 3D modeling, energy and power monitoring, and digital rural revitalization, and promote customers to achieve all-round intelligent upgrading from data production, management to application. At present, we can provide comprehensive solutions, and the planning and implementation delivery capabilities are at the leading level in China.
At present, there are not a few units in the industry that provide remote sensing + AI, but most of them are still in the project establishment and research and development stage, and relatively few can be effectively implemented in national key projects, all because of their extremely high technical difficulties and barriers. Our remote sensing AI intelligent interpretation platform can be said to be one of the few technologies that has been applied in China, and there are quite a lot of application cases.
Since 2018, we have been working on the development of remote sensing interpretation related products, in cooperation with the Shaanxi Bureau of Surveying, Mapping and Geoinformation. After the actual combat verification in the Shaanxi Bureau, V1 was developed0 remote sensing interpretation platform. At that time, the platform could already achieve the integrated processing from data preprocessing to interpretation results, and provide effective automatic auxiliary methods for object classification, target recognition, and change detection in surveying and mapping production. It can also meet the needs of land surveying and mapping, forestry, agriculture, water conservancy and other industries for automatic classification and identification of ground objects, and realize the automatic extraction and rapid update of various resource information.
At present, our remote sensing image intelligent interpretation platform has been updated and iterated to 3The latest version of the interpretation platform has been applied to the natural resources normalization detection project of the Shaanxi Provincial Satellite Application Technology Center of Natural Resources for the entire Qinling Mountains and the whole area of Shaanxi Province. 3.Relying on deep learning algorithms, version 0 of the intelligent interpretation platform can complete intelligent processing such as object classification, change detection and target recognition of multi-source remote sensing images, and provide an integrated and whole-process remote sensing image interpretation solution from data management, sample bank construction, training and interpretation.
Not only in the Shaanxi Bureau, but also many subordinate units of the Heilongjiang Bureau of Surveying, Mapping and Geographic Information also use the platform to assist in improving efficiency and serving the adjustment of crop planting structure in the province; The Quality Inspection Center under the Department of Natural Resources of Ningxia has created an intelligent audit of map products in the whole region based on our platform, which greatly liberates manpower and improves the safety of professional and standardized maps. China Coal Aerial Survey and Remote Sensing Group has incorporated the platform into their overall solution to provide external service ......There are also many such cases.
In terms of remote sensing + AI, we have also built a multi-computing cloud platform and a data storage and management platform, which have been applied in Inner Mongolia Institute of Land and Spatial Planning, Gansu Bureau of Geology and Mining, Gansu Institute of Surveying and Mapping, Shanxi Institute of Surveying, Mapping and Geoinformation, etc., so that they can provide customers with more efficient services.
Reporter: Please introduce the technical development path of remote sensing + AI? YesHow to applyOr a technical bottleneck?
Zhao Hang: In the past, remote sensing, as a technical means of obtaining spatial information data, met the needs of data collection and analysis in the process of digitization. However, the realization of intelligent remote sensing target and scene recognition, information extraction, ground object classification, change monitoring, and three-dimensional reconstruction has always been the focus and difficulty of remote sensing interpretation, and the core key technology of remote sensing + AI is still in the intelligent processing of remote sensing data.
From 2018 to 2019, we mainly processed satellite images, and with the development, we have now been able to interpret and apply the data of UAV flights, and have begun to realize the artificial intelligence processing of pan-remote sensing data. In short, we can interpret, identify and number all graphics, images, ** and other data, and use these functions to implement relevant scenarios and applications. The related applications of UAV flight + AI intelligent interpretation have been implemented in forestry, water conservancy, mining, etc., especially in the safety monitoring and management of open-pit mines, and there are many products and technical solutions.
Since March this year, Chat GPT has been popular in China, making everyone's cognition of large models gradually clear, as an artificial intelligence, in order to achieve a very intelligent and popular application, it is necessary to achieve a large model, at least a large model in local fields. This is also a bottleneck facing the development of remote sensing + AI: only at the level of the Chat GPT large model, remote sensing + AI can achieve more accurate, universal and open applications, but as a small and medium-sized enterprise, it is difficult to achieve such a scale of investment.
But it's not that we're completely powerless in this area, we can start with small scenarios to achieve breakthroughs. At present, we divide our internal algorithms into dozens of major categories of algorithms, and continuously supplement the sample data according to different categories. For example, in the forestry pine wood nematode disease tree identification category, we have accumulated more than 300,000 samples to train the entire model.
The model has been successfully applied to the extraction of pine wood nematode diseased wood in China, and the interpretation efficiency of the whole process is high, from the monthly level to the hour or even minute level. The accuracy and recall rate of intelligent interpretation results have also been greatly improved compared with manual efficiency. Based on this technical idea, we have successfully delivered projects such as the survey of crop planting structure in Heilongjiang Province, the normalized monitoring of water disasters in the whole area of Guangxi Water Resources Department, and the normalized monitoring of the Qinling Mountains in Shaanxi Province.
Reporter: What improvements can intelligent surveying and mapping products bring to the industry?
Zhao Hang: I personally believe that intelligent surveying and mapping is the most important innovative means of new basic surveying and mapping. From the earliest analog surveying and mapping, to digital surveying and mapping, information surveying and mapping, and then to today's new basic surveying and mapping, the surveying and mapping industry still wants to use various new advanced technologies to assist surveying and mapping work and improve production efficiency.
Up to now, surveying and mapping cannot simply be called an industry, but a technology. The goal of the earliest surveying and mapping is to focus on the core responsibility of "two unifications" of natural resources and do a good job in ensuring the "two services and two supports" of surveying and mapping. Nowadays, we will pay more attention to the combination of surveying and mapping and industrial applications, such as surveying and mapping geographic information technology services for agriculture, water conservancy, transportation, energy and land planning. Satellite navigation, drone aerial survey, and lidar are just some of the applications. After collecting data through satellites, drones, lidar and other methods, we can rely on cloud computing, artificial intelligence and other methods to accelerate data production efficiency, and then produce corresponding results to serve the whole society.
With the development in the future, there will be more vivid presentation methods such as digital twins and the metaverse. Its combination with the Internet of Things, from the daily consumption of the map, to the professional aspects of water conservancy, mining, etc., can allow the masses to immerse themselves in its benefits.
I believe that surveying and mapping will eventually develop from an industry, just like the Internet, and become a general technology, which will be integrated into all aspects of people's daily life and management, so that we can be completely inseparable. The final surveying and mapping methods will become more and more abundant, the collection and processing efficiency will become higher and higher, and the presentation methods will become more and more technological, and they will be unconsciously integrated into life and social governance and management.
Since 2016, we have built storage servers, GPU cloud management, and terminal operating machines as the base, and platform services such as organization management, authority management, component management, and resource scheduling as control, and carried out application matrix management through data storage and management platforms, multi-computing cloud platforms, remote sensing AI interpretation platforms, etc., so as to realize data from production, analysis, storage, management to intelligent applications, and form an overall software and hardware solution under the "storage and computing separation" system. Effectively promote the performance and efficiency improvement of intelligent surveying and mapping methods in geographic information, land survey, urban planning, ecological and environmental protection, agriculture, forestry and water conservancy, energy and power, survey and design and other fields, and promote the intelligent upgrading of "surveying and mapping technology" through multi-technology integration.
Recently, the awarding ceremony of the "Southwest Mountain Natural Resources Remote Sensing Monitoring Engineering Technology Innovation Center of the Ministry of Natural Resources" was held. The innovation center will give full play to the advantages of all parties in scientific research, talents, technology and industrialization, focus on the management of natural resources in the southwest mountains and the national strategic needs, carry out innovative research, and strive to solve the technical problems of remote sensing monitoring engineering in the southwest mountains, and create a scientific and technological innovation highland that leads the development of remote sensing monitoring technology and supports the monitoring of natural resources.
The establishment of the innovation center is a new demand for development in the context of digitalization and intelligence, a product of the development of intelligent surveying and mapping, and an extension of the work of the Intelligent Surveying and Mapping Working Committee of the China Society of Surveying and Mapping. The participation of Luxnumber Intelligence shows that our efforts and efforts in the field of intelligent surveying and mapping are gradually emerging. Zhao Hang said that, just like the name "Ress Digital Intelligence", in the future, they will continue to use RS Matrix series products as the starting point to help the industry achieve "digitalization and intelligence".