Data Annotation

How Aya and Oko are Advancing the Next Step in ClimateTech

1. The Client

Oko Environmental is bringing ClimateTech to West Africa. Through the pioneering reforestation methodology of Assisted Natural Regeneration (ANR), Oko is committed to combating climate change through the establishment of indigenous rainforests, growing only native and naturalized tree species in collaboration with local communities.

Their tech-centric approach to reforestation aims to maximize carbon removals whilst boosting biodiversity and supporting communities. By levering multispectral drone imaging, Aya collected real-time data of Oko’s forests housed in an interactive dashboard capable of pinpointing nutrient deficiencies, tracking forest growth, and identifying illegal deforestation activities.By bringing technology to reforestation practices, Oko is surmounting one of the toughest barriers in ClimateTech on the continent: the dearth of pre-existing data on reforestation. Inventing new data points from species matrix, to biodiversity, to success rates, Aya Data is helping Oko lead the way on capturing big-data on regrowing rainforests at scale.

Identifying Systemic Challenges to Build Disruptive AI Solutions

2. The Challenge

Traditional forestry monitoring and analysis are laborious and expensive, often demanding an on-the-ground team manually collecting data, lacking the capacity to capture and analyze big data at scale. Oko was seeking a method to collect data year-round to provide a more comprehensive and accurate picture of their forests’ health condition.

  1. Physical monitoring of sites is time-consuming, subject to human error, and cannot be conducted on a routine basis.
  2. Satellite imagery as a form of data collection is subject to cloud cover and atmospheric interference, which compromises the quality of acquired data.
  3. As reforestation efforts scale, reliable data collection methods for accurate carbon sequestration are necessary for aboveground biomass (AGB) estimation efforts.

Aya’s Solutions team works with Oko to assess the project and find ways to circumvent these challenges characteristic to forestry and climate, a sector historically underserved by technology.

From Acquisition to Annotation: How Vertical Integration Enables Better AI

3. The Solution

Vertical integration represents Aya’s unique capacity to provide three essential services across the AI value chain—data collection, data labeling, and data science. This approach streamlined close collaboration with Oko’s team through every step of the process in order to solve challenges through designing a bespoke AI solution.

Oko wanted to build a model to detect germination rates of trees in several intervals after planting (e.g, 3 months, 6 months) on a 32,000 hectare reforestation project. In order to achieve this;

  1. Drones were deployed in the field onsite in Sierra Leone to collect high resolution RBG data.
  2. Following data acquisition, the drone images were sent to our annotation and quality assurance workforce where bounding boxes labeling was used to detect whether the trees had germinated or not.
  3. This annotated data was used to develop a model that could automatically determine tree germination status when applied to new data. This model achieved an impressive 96% accuracy.
  4. The model was integrated into a dashboard, allowing Oko not only to count the total number of trees in each area, but also to identify spots where germination hadn’t occurred, signaling the need for replanting.

Moving forward, the next phase is to expand the model to identify the different tree species and to estimate the carbon biomass of the trees.

Unlocking Insights With AI to Surmount Climate Challenges

4. The Product

Once production annotation concluded, the dataset was then handed over to our in-house data science team who trained and deployed an AI model for crop identification, germination prediction, tree growth rates, environmental changes and potential risks.

Our machine learning experts present the trained AI model in the form of a customized dashboard available for clients’ individual and continued use. Oko’s personalized dashboard displays data-driven insights in a user-friendly interface.

Chris Rothera, COO of Oko said, “We approached Aya to build a bespoke computer vision solution to monitor seedling germination rates using drones. They developed, trained and deployed the models extremely quickly and with excellent results. I now have a single source of truth dashboard to monitor the success of our reforestation project.”

In Oko’s crop management dashboard, users can click on any tree in the dataset and view key individual insights. Germination rates are tracked via measuring success and failures of tree seedlings. With Aya’s model, Oko has been able to track and understand the success of its forests over time.

View a demo of our dashboard here!

Oko’s story underscores the transformative potential of AI in revolutionizing reforestation strategies, offering scalable solutions to combat deforestation’s environmental impact. Integrating drone imaging and AI technologies not only streamlines tree planting processes but also enables real-time monitoring, fostering data-driven decisions for enhanced forest management.

With continued innovation across AgTech, AI is poised to play a pivotal role in addressing broader climate challenges, promising a sustainable approach to global reforestation and ecosystem restoration efforts.

Machine Learning and AI in Forest Mapping

Our dashboard solutions are bespoke to company, geography, and market needs, and can be adapted to palm, rubber, cocoa, and other forestry and crop management challenges. Future collaboration between Aya and Oko will explore applying drones and AI to tackle broader climate challenges, such as aboveground biomass estimation, which holds immense promise for fostering sustainable and resilient ecosystems on a global scale.

Should you wish to explore how Aya can contribute positively to your project, feel free to arrange a complimentary consultation with our specialists.

The Different Types of Data Annotation Explained

What is Data Classification in Machine Learning?

Leave a comment

Your email address will not be published. Required fields are marked *