DJ Mallmann – AllSource Senior Network Analyst, 40+ years as an imagery analyst
After working in the remote sensing/geospatial industry for the last several decades, I feel well-qualified in declaratively stating that we are in the midst of a satellite, aerial imagery and geospatial data avalanche.
After many trials and tribulations in the industry over the past 20 years, we are finally seeing an exponential increase in the amount, quality and accessibility of relevant overhead imagery in the market. Having extensive imagery coverage and frequent collection revisits dramatically improves our ability to monitor and analyze events and activity, enabling the resulting intelligence to be more accurate. Today, the amount of available coverage is so plentiful, we’re moving away from always having to task satellites (or airplanes and drones) to get what we need. Instead, we can often rely on recently collected archive data found in the commercial imagery providers’ libraries.
With this increased supply in the amount of available imagery data, more companies and investors are focusing their time on developing new algorithms and machine-learning techniques to exploit efficiently this rich and abundant source of geographic information. Many of the most promising new computer-based imagery analytic efforts are focused on statistically identifying and counting objects, such as cars, constructions cranes, or oil tanks. However, a key component of analysis is moving from the superficial identification of objects to truly understanding and qualifying the nuanced patterns that emerge from the imagery. At the risk of sounding a bit like a troglodyte, despite the advancement of computer-based analysis, crowdsourcing and algorithms, I continue to believe strongly in the enduring importance of domain knowledge, tradecraft and expert imagery analysis.
Simply stated, tools without people are simply tools.
Another development in recent years is when the proliferation of imaging data and geospatial data is presented to “the crowd.” I have seen first-hand the potential of crowdsourcing and the positives, such as distributing vast amounts of high-quality, current satellite imagery to the public that can help rapidly identify areas damaged from natural disasters, including floods, fires and earthquakes. I have also seen the risk that can occur when non-experts get involved in imagery analysis, and celebrities become self-proclaimed imagery analysts that can erroneously “identify” missing airplanes in the middle of the Indian Ocean or detect meaningful activity in the middle of Africa or the Middle East.
I assert that to take full advantage of the advantages and capabilities of both computer-aided and crowdsourced image analysis, we continue to need a highly skilled expert community with experience and training in photo interpretation and imagery analysis. This community of experts, many of whom have been professionally trained and worked for years in the remote sensing, defense and intelligence communities can help the crowd (and the algorithms) understand the nuances associated with imagery analysis and the frequent types of errors that occur in computer and crowdsourcing efforts. Perhaps most importantly, this team of domain experts can help scientists create new techniques to improve those methods.
Ultimately, this proliferation of imaging data and geospatial technology is why I believe now is absolutely the right time for AllSource Analysis. Increasingly, there will be problems with clients that are overloaded with data sources, but this points to the company’s core strength: understanding the right information and geospatial intelligence technology and knowing how to apply it with subject matter experts to solve the client’s problems. Having partnerships with many of the imaging and data providers, along with cloud computing and geospatial technology companies, gives AllSource access to a wide variety of geospatial content and solutions and enables analysts to extract the information that quickly solves a client’s difficult question. We hear consistently from clients that they don’t just want an image, they need an answer—and one that cannot simply be counted or statistically measured.