“Rhizomatic” is a software system that uses machine learning and image processing to automatically detect and characterize roots in minirhizotron imagery.
Root observation gives us a wealth of information about the health of plants and their ecosystems. Unsurprisingly, roots can be difficult to study in-place because, well, they’re obscured underground.
Currently, millions of dollars are spent observing plant health from above ground – and for good reason; The Economist recently said, “Over the next 30 years the global food supply needs to increase by 50% to meet the needs of a growing population, even as the food systems carbon footprint needs to be cut in half.”
Current best practices employ minirhizotron imaging–images captured by inserting cameras into small tubes installed in the soil–which allows researchers to capture images over time to track growth. Analyzing the resulting images, however, is incredibly time and labor-intensive and often produces inconsistent, imprecise results. These difficulties not only drain precious time and money, but have discouraged many researchers from using minirhizotrons. They have also made in-situ root monitoring completely infeasible for commercial agriculture.
The ability to trigger crop management events based on root observations would revolutionize global food production.
Oceanit’s Rhizomatic project aims to address this issue by providing a cloud-based AI service to automatically detect and characterize roots from uploaded images, significantly reducing the cost of root analysis while encouraging widespread adoption by researchers in both academic and commercial fields. Oceanit’s approach merges classical computer vision with AI machine learning to create a pipeline that can automatically detect and characterize root systems. Rhizomatic does not require labor-intensive, pixel-level mask creation and instead uses a custom processing pipeline which is able to automatically produce pixel-accurate segmentation.
Rhizomatic’s automation and accuracy has the potential to revolutionize plant studies by lowering costs and removing the main barrier for researchers interested in the field of root science. Scientists have had a good understanding of the visible parts of plants for decades, which led to advancements such as the “Green Revolution” that increased crop yields worldwide, saving entire nations from famine. An improved understanding of roots could have an even greater impact, particularly as we fight environmental degradation and climate change.
Aside from further crop yield increases, there is enormous potential to optimize water use, reduce runoff, prevent over-fertilization, and even measure carbon capture capacity.
The Rhizomatic project is part of Oceanit’s broader effort to develop novel decarbonization technologies at the intersection of plant physiology, soil chemistry, artificial intelligence, and bio-mimicry. Oceanit’s program is referred to internally as, GROOT (Greenhouse Recapture Observation and Optimization Technology).