Satellite Imagery AI: What you need to know
Satellite images are used to solve many business problems and data assymetry challenges. Specifically, industries such as agriculture, geological and hydrological research, forestry, environmental protection, land planning, intelligence and military purposes benefit the most.
In the past, the Earth Observation (EO) practice was majorly available only to government hired analysts, exploring the potential of getting alternative data on what is happening around the world.
There are 3 trends that will make geospatial data more valuable than before. Firstly, satellite and launch manufacturing costs are decreasing. Secondly, small payload launches are cheaper as well. Finally, bandwidth costs have gone down drastically. More SATCOM business gets re-oriented to Observation.
The increasing application of AI transformed the imagery to analytics, allowing businesses and governments to solve challenges, apply location-based services, and do predictive modeling
Where to get data? Open source imagery
The image quality directly depends on the payload and the satellite’s orbit. For example, a resolution of 10 m means that anything below that mark will get pixelated, yet still possible to be analyzed. High resolution imagery ranging from 5 m to 0.3 m become main stream for data collection.
Depending on the business case, :
High resolution with a value of 0.3, m -5 meters.
Medium resolution with a value of 5-10 meters.
Low resolution with a value of 10 meters or more.
High resolution is the most informative , but medium and low resolution are also used in certain cases, specifically on analysis of large land cover.
Commercial high resolution imagery acquisition is not affordable for many businesses. That is why we are sharing these open source datasets open source:
AWS provides open access to a large number of imagery for your query, storing many geo-referenced datasets from commercial and government sources.
If you want to get the satellite imagery of the USA vast agricultral space, you can use NAIP. The data is collected with plane during the growing season.
Facebook created a population data set with an impressive 5-meter resolution for more than 20 countries in Sub-Saharan Africa.
What is it for? Business Use Cases
Identify illegal deforestation. Accessing imagery of the same place over time allows to track changes – crucial to detect illegal deforestation.
For example, satellite imagery analytics technology using already helped Brazilian Amazon rainforest not to be derorestated.
Study refugee settlements. The growing number of displaced people worldwide has led to the construction of new refugee camps and the expansion of existing ones. There is a need to ensure the effective construction and maintenance of these dwellings when it comes to the location of shelters, water sources and other variables. Using satellite imagery can optimize this process.
Help with rescue operations after natural catastrophes such as floods or fires. Satellite imagery analytics allows you to get reliable information about the state of the affected areas of the planet in real time. More than that, in absence of physical access to areas, satellite imagery allows to track the number of people, affected infrastructure and fast response operations.
Economic activity maps . Using the AI model fused with regular survey data, it is possible to estimate per capita consumption expenditures for any place with daytime satellite imagery. This allows you to create fine-grained poverty maps using dwelling type recognition.
Monitor sea traffic. Detecting invisible ships is an extremely important task. Unregistered vessels can perform illegal activities such as trafficking in people and goods.
This allows to optimize, and understand supply chains over the businesses.
The same way satellite imagery can be useful in the case of aircrafts.
Icebergs. The detection of icebergs , which can be a dangerous obstacle to marine activities, has become a much more convenient and effective task with the advent of the possibility of using images from space.
Classic AI Models with Satellite Imagery
Roads detection.
Satellite imagery analytics allows to track the roads, constructions and transport schemas. Roads are easily detectable from above with only 10-30 m resolution. AI models track the network of roads, changes, new constructions. Distance of roads and networks in relations to other assets (roads leading to flooded areas, or post-disaster evacuation, or urban planning) allows to build informative analytics.
Land classifier.
Rivers, lakes, streams, oceans, coastal boundaries, parks, land cover, and beaches can be easily recognized and classified using satellite imagery even at lower resolution. The classic land classifier is the common input to reduce the area of interest in more insightful but requiring commercial high res imagery analytics.
Building footprint.
Automatic building retrieval with the help of satellite imagery can be used in various industries, such as land cover mapping, urban planning, disaster management, and extracting informaiton about human settllement related to other economic activities.
