Machine Intelligence (MI), a term used to describe Artificial Intelligence, Machine Learning and Deep Learning and their application to real world problems, is increasingly being applied to remotely sensed data, especially satellite imagery.
GI specialises in providing a service which adopts new MI technologies to solve specific problems for a range of industries. Our extensive experience in such domains as maritime, transport, agriculture and environment, means that we have the background expertise required to carry out effective MI tasks. Some of our previous projects have included;
- Detecting ships, sea ice, oil spills and monitoring maritime incidents
- Spotting aircraft
- Carbon farming
- Predictive models for the spread of environmental diseases
- Orchard management
- Extraction of complex infrastructure features
MI technology is a cost effective and timely blend of techniques that can be applied to a diverse range of earth observation datasets. GI’s latest Machine Intelligence (MI) vessel detection system can produce detection reports over large, very high-resolution satellite imagery scenes in seconds, while historically this human-based task may have taken hours. Additionally, when MI systems are fused with our other data sources, the results can produce highly reliable, accurate and valuable information.
Machine Intelligence (MI) is a term created to describe Artificial Intelligence (AI), Machine Learning (ML) and Deep Learning (DL) and their application to real world problems. Although these terms are often used interchangeably, they are not quite the same.
AI is a general term applied as far back as 1956, to a computer-based system that possessed some human-like intelligence. Although general AI which is able to understand the world like a human, is still years away, more focussed AI systems that can be used to solve specific problems are now quite common.
ML is an approach applied in the development of AI systems where the machines are programmed to be able to learn and adapt themselves automatically within a framework by giving them access to relevant data.
Deep Learning is another, more recent development in applied Artificial Intelligence. Deep Learning uses advances in computer technology to train deep neural networks (including convolutional networks) in a way similar to how a biological brain functions.
Machine Intelligence (MI), groups these techniques together when focussing their application to real world problems.
For more information about the history of these techniques the computing hardware manufacture Nvidia has a summary (https://blogs.nvidia.com/blog/2016/07/29/whats-difference-artificial-intelligence-machine-learning-deep-learning-ai/)
Utilising MI is not only a computationally efficient choice of technology, but it is also a highly effective way of applying algorithms to a wide and diverse range of environments. An advantage of MI systems is that they allow temporal, spatial and dimensional data fusion. For example, an MI system could fuse optical and SAR imagery across multiple capture opportunities as part of its automated analysis.
MI systems are heavily dependent on high-quality training datasets. These datasets can either be gathered manually, through a trusted source, or developed using a combination of remote sensing methods. An alternative to these methods is open source intelligence (OSINT). GI has extensive experience in developing tools to geo-reference unstructured OSINT data for this purpose and have effectively applied these techniques in a variety of scenarios.
Open Source Intelligence
At Geospatial Intelligence, we have expertise in gathering open source intelligence from a broad range of sources using customised and developed software tools to analyse, geo-reference and spatially understand what is happening, and where, for our clients. Our team can also customise processes and software to fit with customer requirements.
Working to solve our customer’s problems, we use a range of artificial intelligence (AI), machine learning (ML), natural language processing (NLP) and data analytics techniques to process datasets ranging from a few hundred, to millions of data points.
Open source intelligence (OSINT) is intelligence material produced from legally procured information from the public domain. The information can range from social media posts through to news articles, government reports, videos and images. OSINT can be particularly valuable due to the shorter timeline required to acquire and process it compared to more traditional intelligence sources. Geospatial information can often be derived from OSINT sources and commonly big data mining techniques are used to derive intelligence from the information. OSINT can be applied to a range of fields, including cyber security, financial analysis, health and pharmaceuticals, insider threat, fraud detection and national security.