By: Tyler Foreman
Enterprises have understood the value that lies locked within geovisual data for many years. Advances in remote sensing technologies have made the collection of this data more cost-efficient. At the same time, enterprises are under increasing pressure to proactively monitor and manage risk, driving the need for more frequent visual assessment. These factors are driving a rapid increase in geovisual data collection.
As the volume of data has increased exponentially, the challenges of managing these data types have been difficult to overcome with traditional data management approaches and platforms. As a result, most of the data collected is relegated to siloed network shares or USB hard drives that few in the enterprise even have access to.
What is geovisual data and what is the opportunity it presents?
Generally speaking, geovisual data is collected by remote sensing devices such as drones, satellites, manned aircraft and autonomous vehicles. The output of these devices produce data that is visual – RGB imagery, thermal imagery and 3D point clouds, among many others. The data is also geotagged – that is, it contains metadata that describes the exact (with varying levels of accuracy) geographic location where it was captured.
Geovisual Intelligence is business intelligence that is extracted from this data. This intelligence reveals the dynamic reality of critical assets and infrastructure. That is, it provides the ability for near-time, in-situ visual analysis that can answer the following questions:
- Where is my asset located?
- What is attached to/housed by it?
- What condition is it in?
- What is around it?
Answering these questions is at the core of many critical workflows for different industries including risk assessment and conflation of asset records in enterprise data systems. When coupled with automated analytics and AI, geovisual data can answer these questions and transform these traditionally manual and inefficient workflows to be automated and run at a high frequency and scale.
What are the challenges extracting geovisual intelligence?
What makes extracting geovisual intelligence difficult to do at scale boils down to data management challenges. Geovisual data such as RGB imagery, LiDAR point clouds, geoTIFF orthomosaics and others can typically be characterized as large binary files that often need to be maintained as groups of related files or datasets (ie: a drone survey consisting of many RGB images). Efficiently moving and organizing these very large files and datasets to a centralized location and making them easily accessible for viewing/analyzing can be very challenging and cumbersome with traditional approaches.
Even when the data is accessible to users, the visual nature of the data makes manual analysis extremely tedious, unreliable and inefficient at scale. Relying on human operators to sift through and analyze these datasets is simply not practical. The results are often highly subjective and the quality can be affected by factors such as fatigue, attention to detail and depth of institutional knowledge held by the human operator.
Transforming geovisual data into geovisual intelligence
Rapid advancements in AI has created new capabilities in computer vision and visual data analysis that was unthinkable just a few years ago. Deep Learning techniques in particular are well suited for analyzing massive amounts of geovisual data and have rapidly matured into commercial grade analytics.
In order for these analytics to provide true value to an enterprise, they must be deployed within an analysis pipeline that connects a robust geovisual data management platform with a configurable workflow orchestrator that can execute a sequence of analytics against the right data set and deliver the intelligence extracted from that data directly to enterprise systems downstream.
The Intel Insight Platform provides this foundational geovisual data management and workflow orchestration platform. At the core of the platform is the Intel Geovisual Data Lake, which provides services specifically designed to solve the challenges of managing geovisual data at scale. This includes a suite of services for data ingestion, storage, cataloging and preparation. Designed for high availability, reliability and security, these services are designed to meet the challenges of managing geovisual data and extracting business intelligence at scale.
When coupled with the Intel AI Workflow Orchestrator, geovisual intelligence can be automatically extracted using AI developed by Intel, third party partners or even customers and connected directly to enterprise systems. This end to end workflow capability provides enterprises with an open framework for realizing the benefits locked within their geovisual data.