HomeMioTech ResearchArticle Detail

How Can Remote Sensing Data and AI Help Financial Services

In China, AI solutions based on remote sensing data are being tailored for the finance industry on investment research and decision-making.

Tracey Xiang2019-05-22

Remote sensing data collected by satellites and other flying machines as an alternative data source is increasingly adopted around the world. The availability of the increasing high-quality data aside, the significant advances in artificial intelligence, deep learning in particular, and increases in computer processing power achieved in recent years have considerably increased the usability of such type of data with much improved accuracy of object recognition and predictive analysis capabilities.

In China, the finance industry is one of the major target markets for remote sensing data and related application developers.

Noise Reduction of Satellite Imagery. Source: MioTech

AI-powered services based on or has factored in remote sensing data are being developed in China to help finance companies to make credit decisions, insurers in damage evaluation or pricing, or traders on commodities research. "With the help of AI applications, not only can we identify related entities within a frame over time, but more importantly, identify patterns, outliers to not only verify models or research, but also predict and forecast future activities", according to MioTech, an AI data analytics service provider targeting financial services.

The Chinese government has been a major player in satellite manufacturing and launch. Riding the commercialization trend of the space industry, a wave of privately-owned startups have established in China since the regulatory green-light was given in late 2014, exploring opportunities in lower-cost manufacturing and launch of small and micro satellites. Chinese satellite data providers or application developers also acquire data or services from foreign companies like the Planet Labs.

Low altitude remote sensing technologies, such as airborne imaging systems and sensors flown on drones or manned aircraft, and advanced data acquisition techniques are also increasingly adopted in China.

More than just a new data source, remote sensing data are considered especially valuable as a reliable source for tracking and analyzing the economic activities of China, the world’s second largest economy and the second largest country by land area, as China’s official economic data have long been suspected to be unreliable given the increasingly revealed data manipulation scandals and alleged deficiencies in the methodology used by China’s state statistics bureau. According to MioTech, remote sensing data are able to provide "concrete, tangible large scale oversight when identifying, verifying, analyzing and predicting economic activities at a specific location." This type of data and related services can also be timelier than the conventional state-provided economic metrics.

Source: SpaceKnow

SpaceKnow, a U.S.-based AI company that monitors economic activities based on satellite imagery, developed Satellite Manufacturing Index (SMI) to reflect and predict industrial activities in China. The SMI tracks satellite imagery of over 6,000 industrial sites across China. Apart from a higher frequency of releases than the official manufacturing Purchasing Managers Index (PMI), the company claims the SMI outperforms the official PMI in modeling the current and future growth of trade volume.

Remote sensing data and AI are especially valuable for understanding China’s agriculture sector. Given the lack of basic data and shortage of adequate risk modeling, rural and agricultural finance and insurance in China are quite backward and have been considered highly risky. As China is heavily promoting efforts to boost rural economic development to close the gap between urban and rural areas, it is believed that the capabilities for performance monitoring and predictive analytics that are enabled with remote sensing data and AI will be revolutionary for the rural and agricultural finance and insurance sectors in China.

Thanks to the availability of lower-cost, easy-to-use unmanned aerial vehicles (UAV), automatically identifying and counting crops or animals, measuring acreage and real-time monitoring have become much easier to come by.

AI techniques help increase the accuracy of the recognition and classification of crops or animals, make forecasts on harvests, or detect anomalies. AI-powered applications tailored for financial companies are thus able to help lenders or insurers price risk based on the health of the agricultural products or production forecasts, manage risks by providing monitoring services or alerting them whenever anomaly happens, or evaluate damages.

“Harvests no longer need to be at the whim of nature. Analyze nature for your havests.” Source: GagoGroup

China has been the world’s largest producer and consumer of pork. China-based remote sensing startup GagoGroup provides financial companies a platform that can monitor pig farms, assess the financial health of pigs, and evaluate death loss. The company has patented a computer vision AI technique in China that estimates the body length and weight of pigs with an average accuracy of 96%. (link in Chinese)

Chinese remote sensing startup TerraQuanta provides private equity firms and traders with remotely sensed data and insights about a wide range of agricultural crops.

Apart from agricultural commodities and industrial productions, Chinese remote sensing data startups are also developing solutions for other usage scenarios such as evaluating business performance of malls or parks by monitoring and analyzing car traffic. “Not only are we able to identify the traffic in that area, we can also overlay on top of the data, information on transaction volume at nearby POS (points of sale) machines or in terms of reception how cell phones are connected to specific signal point. So overlaying all these various layers over a geospatial data can help investors gain insights on population density and activity at that specific point in time", as commented by MioTech.

Major computer vision AI developers have also started building solutions tailored for remote sensing data. SenseRemote, the deep learning solution for remote sensing imagery developed by one of the world’s highest valued AI startup SenseTime, is fully automated, has a recognition precision rate of over 90%, and enhanced processing speed.

Some studies have shown that AI solutions tailored for remote sensing data can replace human in plant and animal counting as they have a much higher accuracy than ground or human-made observations, and they are much faster and less costly. When it comes to investment decision-making, remote sensing data and related services is still only a complementary source of information. “It helps reduce risk when it comes to making decisions. Financial reports, currently in terms of predicting power, is still the most helpful. But satellite imagery gives you further clarity, greater veracity to your research", according to MioTech.

Tracey Xiang is a tech writer, specializing in ChinaTech, Digital Economy, FinTech and AI.