The ICLR 2020 Workshop on Computer Vision for Agriculture took place virtually on April 26, 2020 featuring talks from both NASA Harvest Africa Lead Dr. Catherine Nakelembe and NASA Harvest machine learning specialist and Domestic Lead Dr. Hannah Kerner, with contributions from NASA Harvest Program Director Dr. Inbal Becker-Reshef. The workshop agenda centered around the reality that artificial intelligence (AI) is becoming ever more prominent in the agricultural industry due to rapid technology advancement in the past several years. The vast range of topics discussed by workshop presenters and panelists included drone automation for crop monitoring, the improved technology of agricultural equipment, satellites and remote sensing for monitoring food security indicators, the use of apps to connect farmers to data and vice versa, and many other aspects of technology that influence agriculture today. The workshop was coordinated through the joint efforts of AI and computational agriculture researchers with the support of CGIAR and the Radiant Earth Foundation. All presentation slides and full video playback of the workshop sessions are available online.

Integrating Earth Observations in National Agriculture (Dr. Nakalembe)

For many African countries, food security is one of, if not the most, pressing issues of today. My job as Africa Program Lead for NASA Harvest is to help countries build their own agricultural monitoring systems based on free and low-cost satellite data to inform life-saving decisions related to food security sooner and with a deeper evidence-base. Our work at NASA Harvest focuses on using satellite data to understand agriculture and food security at local to global scales. Satellite data have been used in agriculture for decades by organizations like USDA, but not in many other countries especially in Africa. NASA’s Applied sciences programs Harvest and SERVIR are redoubling NASA’s investment in making satellite data more useful for agriculture monitoring. Combined with field data, satellite data help us understand sooner, and at a larger scale, what food crops are growing where, how they are doing, estimate how much food will be produced in a season, and provide early warnings of crop failure. This early warning gives organizations more lead time to prepare responses and even mitigate impacts altogether.

Field-Level Crop Type Classification with k Nearest Neighbors: A Baseline for a New Kenya Smallholder Dataset (Dr. Kerner, Dr. Nakalembe, and Dr. Inbal Becker-Reshef):

Accurate crop type maps provide critical information for ensuring food security, yet there has been limited research on crop type classification for smallholder agriculture, particularly in sub-Saharan Africa where risk of food insecurity is highest. Publicly-available ground-truth data such as the newly-released training dataset of crop types in Kenya (Radiant MLHub) are catalyzing this research, but it is important to understand the context of when, where, and how these datasets were obtained when evaluating classification performance and using them as a benchmark across methods. In this paper, we provide context for the new western Kenya dataset which was collected during an atypical 2019 main growing season and demonstrate classification accuracy up to 64% for maize and 70% for cassava using k Nearest Neighbors—a fast, interpretable, and scalable method that can serve as a baseline for future work.