Over the past year, thanks to the rapid development of AI technology, the processing of remote sensing data has become more intelligent and automated. The AI remote sensing sector occasionally announces financing news, and a trend of industry warming is gradually emerging.
According to incomplete statistics, in 2023, the global vegetation intelligence platform Overstory completed a Series A financing round of $14 million; AI remote sensing company Impact Observatory completed a seed round of financing of $5.9 million; the optical and radar imaging satellite manufacturing service provider Zhongke Xingrui Technology completed a Series A financing round of over 100 million yuan; and AI remote sensing application company LiveEO completed a financing round of $10 million...
In the new year, the AI remote sensing industry continues to welcome more innovation and development. In January, Jiahe Technology, one of the earliest remote sensing agricultural startups in China, announced the completion of nearly 100 million yuan in Series B+ financing. The company plans to use this round of financing to promote applications for agricultural and insurance customers, continuously acquiring customers through mature solutions.
It is clear that with the continuous breakthroughs in AI computing power and the maturation of algorithmic models, the application of "AI + remote sensing" is becoming more and more widespread and is full of infinite possibilities.
Advertisement
The AI Dream in the Niche Track
Some say that remote sensing technology is like "giving the Earth an extra pair of eyes," a phrase that vividly describes the role and importance of remote sensing technology.
Remote sensing technology acquires electromagnetic radiation information from the Earth's surface, obtaining image data and other related data from various places on Earth. By analyzing this data, information about surface objects, topography, and meteorological conditions can be obtained. On this basis, remote sensing technology is applied in multiple fields, such as environmental monitoring, agriculture, urban planning, resource management, and disaster warning.
Remote sensing technology is of great significance to the digital development of traditional industries, but compared to other popular technology fields, it is considered a niche track. The "niche" aspect lies in the relatively low attention and popularity of remote sensing technology, which is due to some challenges in its application.
Firstly, under traditional manual or low-automation modes, there is an insufficient supply of data processing and interpretation capabilities. As technology develops, the data information contained in remote sensing images becomes increasingly rich. Traditional manual interpretation requires a lot of time and professional knowledge, and the efficiency of processing large amounts of remote sensing data is low, making it difficult to meet the industry's demand for rapid processing and agile application of massive data.Secondly, the integration and sharing of remote sensing data has always been a challenge. The multi-source data used in the remote sensing field include optical, radar, infrared, nightlight, etc., which come from different sensors. Each sensor has its specific working principles and measurement capabilities, so integrating these data at the level of fusion is somewhat difficult.
Thirdly, the acquisition of remote sensing data requires a large amount of equipment, and the investment cost is high. As we all know, satellites and aircraft are the main platforms for obtaining remote sensing data, which require expensive research and development, manufacturing, and launch costs. Especially for high-resolution satellites, they also need to be equipped with advanced sensors and equipment, and their costs are very high.
With the acceleration of the digitalization process in the industry, market demands are becoming more and more diversified, bringing more challenges and pressures to remote sensing companies. To cope with these challenges, remote sensing companies are deeply cultivating AI remote sensing technology, trying to use artificial intelligence technology to improve the processing and analysis capabilities of remote sensing data to meet the diversification of market demands.
In summary, the value of AI processing remote sensing data is gradually being recognized, and the integration of AI and the remote sensing industry will become more and more in-depth. Especially with the explosion of AI large models, companies such as SenseTime and Jiahe Technology have successively launched AI remote sensing large models and AI remote sensing services, and the competition of the new round of AI remote sensing is more intense.
SenseTime "Encircling" Remote Sensing Large Model
Last year, SenseTime's stock was reduced, the founder Tang Xiaoou unfortunately passed away, and it was shorted by the American short-selling agency Grizzly, which was a roller coaster. Various unfavorable news reflected on the financial report, turning into a double decline in revenue and net profit. SenseTime's financial report shows that in 2023, the company's revenue was 3.406 billion yuan, a year-on-year decrease of 10.57%, and the net loss was 6.44 billion yuan, a year-on-year decrease of 6.54%.
Despite SenseTime's poor overall performance in the face of fierce market competition, there are still some aspects that perform well, such as generative AI business. SenseTime's financial report shows that in 2023, the revenue of generative AI reached 1.2 billion yuan, accounting for 35% of the group's total revenue, achieving nearly a 200% increase.
Since the commercial application of AI large models has been launched, SenseTime has continuously expanded the commercial application map of AI large models, trying to "encircle more land and occupy more positions" as much as possible, striving to seize more market share. Against this background, the remote sensing large model "SenseTime Boundary" came into being.
In May 2023, SenseTime launched the "SenseTime Boundary" (SenseEarth) intelligent remote sensing analysis and geographic information application cloud platform. The remote sensing large model "SenseTime Boundary" can analyze remote sensing data of any time and resolution within China, with more than 40 types of surface object segmentation capabilities and industry-leading remote sensing analysis performance.
As a company with AI genes and a deep understanding of cutting-edge scientific and technological fields such as computer vision and deep learning, SenseTime is committed to promoting the integration of AI and the remote sensing industry, and is expected to form a new type of remote sensing driven by technology, thereby obtaining new growth.On the one hand, SenseTime, based on its general AI infrastructure SenseCore and the "daily renewal" large model system, has built a comprehensive generative AI product system, allowing it to take a step ahead in the "AI+Remote Sensing" field. The application of these technologies and systems enables SenseTime to provide higher quality and more intelligent remote sensing solutions, thereby winning the favor of industry users.
So far, the AI remote sensing large model has served more than 20,000 industry users, covering many fields such as natural resources, agriculture, finance, environmental protection, photovoltaics, etc. Especially in the field of natural resources, it has been applied in the law enforcement and supervision of natural resources in more than 14 provinces and cities, increasing users' work efficiency by 3-5 times.
On the other hand, SenseTime cooperates with various big data suppliers and satellite manufacturers to comprehensively upgrade the platform's data sources, build a strong differentiation and cost-effective super data analysis service, and further enhance product competitiveness.
SenseTime cooperates with the three major domestic remote sensing data platforms "Jilin No.1 Network", "Four-Dimensional Earth", and "Star Map Earth". These platforms have a large amount of remote sensing data resources and rich geographic information data. SenseTime can integrate data resources to further enrich and expand its own remote sensing database.
As the saying goes: "Technological innovation is the driving force of development". SenseTime, with its strong AI foundation and a variety of remote sensing data provided by partners, has created a remote sensing large model that may open up new growth space for SenseTime.
Jiahe "Plays with" AI Remote Sensing Agriculture
Some people say: "Farmers are shifting from traditional weather-dependent farming to relying on artificial intelligence for decision-making and agricultural production", which makes sense. With the development of technology, the integration of artificial intelligence in the field of agriculture is getting deeper and deeper, and its application in agriculture is becoming more and more extensive. Some enterprises focusing on AI remote sensing agriculture have made significant progress, such as Jiahe Technology.
After years of development, Jiahe Technology has gradually transformed from an explorer of AI remote sensing agriculture to a leader. It has not only developed the Remote Sensing Intelligence Computing Platform and the Remote Sensing Data Platform, but also successfully cooperated with more than 2,700 customers and piloted many commercial directions.
It can be said that with the support of AI remote sensing, Jiahe Technology has opened up a new growth space. Under the exploration of Jiahe Technology, AI remote sensing technology has also achieved wider application and higher efficiency.
In terms of research and development, Jiahe Technology cooperates with top remote sensing research institutions and has accumulated profound AI algorithm technology and data application experience. Currently, Jiahe Technology has cooperated with more than 20 top remote sensing research institutions at home and abroad, such as Wuhan University, Chinese Academy of Agricultural Sciences, and National Meteorological Administration Satellite Application Center, and has jointly carried out the design and research and development of more than 40 AI special industrial research projects.Jiahe Technology continuously introduces cutting-edge AI algorithm models into AI remote sensing agriculture, enhancing the efficiency of remote sensing data analysis applications, and providing customers with remote sensing data services that are high-precision, cost-effective, and of high scenario value, which also demonstrates its own technical strength. It is understood that Jiahe Technology, with its self-developed algorithms and Luojianet as the core, deeply integrates with the Ascend MindSpore AI framework, reducing remote sensing data processing time by 50% and saving costs by about 30%.
In terms of application, Jiahe Technology introduces advanced AI algorithm models to enhance the intelligent level of AI remote sensing agriculture, improve the convenience and timeliness of user data access, and also explore broader business fields and commercial opportunities for itself. For example, facing agricultural planting enterprises, agricultural input enterprises, agricultural financial enterprises, and agricultural and rural bureaus, Jiahe has launched two SaaS software, Diandian Tian and Nongxiantong, providing various data services such as plot segmentation, crop identification, growth monitoring, disaster assessment, disaster early warning, and yield estimation, allowing users to achieve "one-click query" of crop production.
According to official information, Jiahe Technology's database already has data models for more than 20 major crops, achieving satellite data service coverage for agricultural plots across the country, and can even provide overseas services. The number of customers has reached more than 2,000, and the revenue has been growing at an annual rate of 50%.
Today, Jiahe Technology has grown into a high-tech company based on agriculture. As Jiahe Technology continues to strengthen the penetration of AI remote sensing into the industry, it is expected to make greater strides in a broader field in the future, further opening up the imagination space, and continuously increasing long-term value.
Large model empowerment, AI remote sensing accelerates again
With the advancement of remote sensing technology and the improvement of data acquisition capabilities, people can obtain more and higher resolution remote sensing information. The innovation of "AI+remote sensing" applications urgently needs to accelerate, and the emergence of large models just fills the gap in this demand.
On the one hand, with the continuous development of remote sensing technology and the improvement of data acquisition capabilities, remote sensing companies have generally recognized the importance of large models in applications, and the remote sensing industry embracing large models has become an industry consensus.
Simply put, the massive data of the remote sensing industry provides opportunities for upgrading large models, and the strong data training capabilities of large models will promote the development and innovation of the remote sensing industry. The Taibo Research Institute's "AI Remote Sensing Large Model Market Research Report (2023)" shows that by 2025, the market size of AI remote sensing large models will exceed 20 billion yuan.
On the other hand, AI companies, Internet companies, entrepreneurs, and others have successively entered the field of AI remote sensing, and the overt and covert struggles of people from all walks of life will become the driving force for the advancement of the AI remote sensing industry. It is believed that with the continuous progress of AI technology and the improvement of hardware computing power, the application of large models in the field of remote sensing will become more and more popular, and will further promote the development of AI remote sensing.Taking Alibaba as an example, in October 2023, Alibaba's DAMO Academy today released the industry's first remote sensing AI large model (AIE-SEG). This model can recognize nearly a hundred types of remote sensing land cover classifications and can also automatically optimize recognition results based on interactive user feedback. In some specific scenarios, compared to traditional remote sensing models, the accuracy of instance extraction can be increased by 25%, and the accuracy of change detection can be increased by 30%.
It is not difficult to see that the field of remote sensing requires large models. In the future, remote sensing technology will become more intelligent, automated, and refined, providing more convenient and efficient services for people's production and life, and will also bring new development opportunities and challenges to related companies such as SenseTime Technology and Jiahe Technology.