Improving productivity in pea, bean and legume growing through advanced data analytics, machine learning and artificial intelligence techniques

"Each year the UK imports \>3Million Tonnes of soya protein for use in the pig, poultry and dairy sectors, 98% from South America (the growth of soy plantations being a key factor in the forced eviction of many rural communities, excessive pesticide use and deforestation).

It is widely recognised that the UK could cut its dependency on imported soya by \>50% by encouraging farmers to switch to home-grown alternative protein crops notably pulses - field beans and peas. However, yield instability, slow rate of germination, challenges in crop management including soil health and lack of confidence in gross margins, have limited efforts to increase home grown protein production.

The potential for increased UK production in pulses is higher than in any other crop, however this can only be achieved by creating greater confidence across growers for the inclusion of legumes into crop systems. Improved understanding of factors impacting yield performance and field variability are critical, and whilst precision agriculture techniques such as the use of satellite imagery and ground sensors for plant health data capture are being trialled across wider crops, no form of remote sensing exists specifically to accommodate the crop physiology and growing requirements for pulses with none truly capable of direct application.

The partners (Hummingbird and PGRO whose members account for 80% of the UKs legume market) aim to address this market gap through the development a solution which uniquely combines UAV collected imagery along with handheld and historical data with a proprietary data analysis and Artificial Intelligence based crop monitoring platform specific to pulses, to provide real-time actionable intelligence across the crop cycle to give an accurate depiction of in-field performance. Data generated will support growing decisions from varying seed rates through to weed classification, disease detection and accurate prediction of crop prior to harvesting, all aimed at improving crop management, enhancing yields/quality and growth confidence. To date, concept design has been completed with IUK support now required to develop the platform into a proof of concept prototype and based on two seasons of data collection and system testing. If successful, the solution will represent the first remote sensing software analytics for peas and beans growers with the potential to double yields, optimise inputs and gross margins, and alleviating the environmental burden of sustainable intensification of agriculture. The project will deliver significant export led growth for Hummingbird, increased employment with future R&D application across wider crops."