Planet’s bringing down terabytes of imagery every day. With an archive this large, spotting change needs to be fast and accurate. Conducting manual analysis can provide reliable results, but requires tedious and time-intensive image interpretation. To produce change insights at speed and scale, automation is needed.
But what constitutes change? Any two images of the same area will have small variations, which might be the type of change you’re looking for. Due to the way computers interpret pixel data – or in a sense, ‘see’ images – automated change detection is prone to false positives. For example, differing weather conditions, the time of the year in which imagery is captured, and differences in acquisition angles can make buildings or scenery appear differently in images of the same location. For true automated change detection to work, it must compensate for all of the factors that can produce these false positives.
Exogenesis, an Earth observation and automation analysis company has been working with Planet imagery and the Planet API to solve this problem.
“We have developed a unique change detection tool called Mandala-EI that uses sophisticated algorithms to solve the false positives problem while providing full automation of the process,” says Sean Anklam, co-founder and Executive Director at Exogenesis.
Two Planet satellite images of Manhattan taken in November 2016 (left) February 2017 (right). Because of differences in sunlight, conditions can create false positives of change detected by computers.
Armed with Planet imagery, Exogenesis is working to reduce the colossal amount of time and manual effort it would have taken to process and discern meaningful change over large areas of interest. With Planet’s high cadence satellite coverage and data delivery via the Planet API, Exogenesis is interpreting change more efficiently and accurately. In some cases, Mandala-El has reduced a task of many months to less than three days through automated vision methods.
The above images show Santa Barbara before (top) and after (bottom) severe rain. Change detection tools from Mandala-EI (right) can classify severity of changes between the images. Here, pink represents significant change, yellow shows moderate change and green identifies false change.
Partnering with Exogenesis helped us deliver highly accurate change detection and imagery analysis quickly and on a massive scale. But the promise and potential of the joint Planet and Exogenesis solution is not limited only to this project. Your industry may be one of many that takes advantage of satellite imagery and requires a faster, more accurate way to perform change detection and analysis. Planet’s persistent delivery of earth imaging paired with powerful automation tools like Mandala-EI is moving global change detection towards real time.
Building your own geospatial data solution? Learn more about Planet’s Application Developer Program.
To read more about the project between Planet and Exogenesis, download the full story.