GEO-AI: A Mapping Gamechanger Today

 

WHAT IS GEO-AI?

The twenty first century has been characterized by rapid technological changes that are a build-up to the insights of scholars and scientists of the past century. The geospatial world is not an exception with the rapid demand for location-based solutions fueling the paradigm shift. In particular, Geo-AI is a technological trend to watch out after taking the industry by storm with the internet being a catalyst to its recent steady growth. Before sinking any further into the subject, what actually is Geo-AI?

Geospatial Artificial Intelligence (Geo-AI), can simply be stated as a concept of automation whose intent is to facilitate fast decision-making and pattern recognition of geographic data through subsets like deep learning which entails the use of neural networks just as the human brain. The data is trained before being executed using a deep learning model. Despite the introduction of the term AI in the mid 1950’s by Stanford University’s professor John McCarthy, its growth has been witnessed with the emergence of big data due to the accelerated use of the internet in the past decades.

Figure 1. Relationship between artificial intelligence, machine learning and deep learning

Source: Geojobe

Geo-AI is a timely technology as the world heads to the fourth industrial revolution that is keen on optimization of information; with practical examples including change detection, landcover classification, impervious mapping and compliance monitoring of physical projects.

 

HOW GEO-AI IS A GAMECHANGER

Mapping has evolved over thousands of years ranging from cave paintings over two millennia ago to paper maps and of recent past digital desktop and web mapping. A scientific discipline that is proving to be essential in almost all human endeavors, the limitations that come along with its use on matters data acquisition, processing, storage and dissemination are sought to be eliminated as time goes by. Just in the recent past, remote sensing images were captured by tagging cameras on flying birds. As an improvement, the use of low-flying planes and then satellites have been used for data acquisition.

 Digitization of data on images would entail physical drawing and later on the use of mouse on a digital screen. This tedious engagement compromised the efficiency and effectiveness of the output for objective decision making. The reason is that humans, who are at the center of the process were bound to be biased due to the repetitive nature of the work that entailed huge datasets. The contemporary inculcating of computing into the fast-evolving discipline has seen increased productivity and enhanced efficiency throughout the value chain and therefore professionals focus on the important tasks.

Geo-AI and data mining have enabled object-detection, clustering and classification of high-quality images in the remote sensing domain (technological advancements however aim to accommodate low-resolution images). This has been facilitated by deep learning which is used when spatial data does not conform to a specific structure. Big data infrastructures like Hadoop and open-source machine learning platforms like Geopandas for geospatial analysis are a positive force for the sustainability of Geo-AI and its use across disciplines.

 

Figure 2. Extraction of features on a remote sensing image using GeoAI.

Source: Multidisciplinary Digital Publishing Institute (MDPI).

 

The use of Geo-AI is being seen in detecting terrains, extracting of information from scanned maps as well as in spatial interpolation. Machine learning and natural languages are used in geospatial artificial intelligence in acquiring spatial information from unstructured data like texts. Acquired user data through mobile sensors and IoT-enabled devices helps in understanding of human mobility patterns through Geo-AI in order to provide planning solutions in the context of urban areas; a concept known as social sensing. To bring the point closer home as in the case of the Covid-19 pandemic, the use of Geo-AI has been heavily used in the disease modelling and prediction as well as surveillance. It has been possible with the help of maps through analysis of geotagged social media data.

 

OVERCOMING THE LIMITATIONS TO GEO-AI ADOPTION

Needless to say, there have been dissenting opinions regarding the emergence and success of geospatial AI. These range from privacy and security concerns pertaining data mining, limited projects for Geo-AI application because of unawareness and biases in data. More so, there is a wide range of temporal resolutions for geodata as well as financial and technical constraints for countries in the global south. Nonetheless, it is important to note that Geo-AI is a technological trend whose work is in progress.

 Moreover, the fear of being overtaken by the new technology limits sight of opportunities and it is important that the intended consumers improve their skills to be competitive in the evolving world. Technology transfer and public-private partnerships would also be of essence in ensuring that these geospatial developments are embraced globally.

WHAT’S NEXT FOR GEO-AI?

To conclude, it is high time that stakeholders in the geospatial discipline embrace geospatial artificial intelligence through promoting awareness and facilitating research that would inform future developments in the industry including solving global climate change issues. Geo-AI as a form of automation is here to stay and at the opportune moment when there is need to build capacity for spatial big data integration and analysis; a field that is predicted to aid in reducing the geospatial digital divide in a decade’s time.