Future food security with algorithms and drones

An algorithm created by two Mohamed Bin Zayed University of Artificial Intelligence (MBZUAI) students could help revolutionize the country’s agriculture industry while reducing waste. MBZUAI Masters students Mugariya Farooq and Sarah Al Barri created a machine learning framework which can classify plant diseases from an image and also predict yield using basic data inputs. Their work earned them second place at the Agritech Hackathon or Agrithon, organized by the Abu Dhabi Agriculture and Food Security Authority (ADAFSA), and held as part of the Abu Dhabi Agriculture and Food Security Week in November.

Farooq said they were given two days to complete their project on either food security or biosecurity. “Some diseases are communicable for plants as well,” she said. “So our work tries to remove that plant at an early stage to reduce waste. It’s all about food security. How do we secure the supply? How do we reduce waste? When you see the results, you see the model predicting.”

Farooq said business impact in reducing resources and waste was key to the success of their project. “You don’t have a research scientist available everywhere or an agricultural scientist available which can go through all the rows of plants and say this one has disease,” she said. “We suggested the usage of drones. While the drones are taking pictures, we can get it in our system and make predictions. We also tested the accuracy of our algorithm so if an agricultural scientist saw the same plant leaf and our algorithm saw the same plant leaf; how far away are they from each other? It was quite accurate (above 99 per cent).”

Billed as a hackathon, the Agrithon was launched to tackle challenges facing the food security sector. ADAFSA’s goal is to tackle food-security challenges through collaboration, innovation, and the pushing of boundaries to identify agri-tech solutions. Al Barri said using the plant disease and the yield predictions, farmers would know how much to grow. “We don’t want to over grow, and have it end up in the supermarket being thrown out,” she said. “Given some parameters like the temperature, how much water the soil is getting, what type of soil it is and everything, we can predict the yield so we can actually have a better supply chain. “At the same time, if a plant has a disease, we wouldn’t have to waste water or resources on it.”

The Agrithon was held for the first time in Abu Dhabi and is one of the main events of Abu Dhabi Agriculture and Food Security Week, the largest sustainable agriculture, food security and biosecurity event in the Middle East and North Africa (MENA) region. The pair were awarded 30,000 AED for their hard work at the hackathon. ADAFSA authorities are interested in further developing the student’s work and looking at possible development for applications in the field. Farooq and Al Barri will complete their Masters in Machine Learning in 2023. Fellow MBZUAI team, Masdar Boys, won first prize for developing a dashboard, which integrates multiple machine learning models for visual plant disease diagnosis, optimal animal clinic placement, and disease outbreak zone classification.