Crop waste poses significant challenges in orchard management, impacting both economic returns for growers and the efficiency of the entire supply chain. Accurate yield estimation is crucial in mitigating these issues, and advancements in Machine Learning (ML) and Artificial Intelligence (AI) have demonstrated superior performance over traditional human methods.
“Crop waste is a major factor to consider in managing orchard’s efficiently and sustainably, reducing waste has so many benefits to growers and to the whole supply chain.
Accuracy of yield estimation is fundamental and this technology has demonstrated that Machine Learning and AI is way ahead of a human!” ~Johnathan Sutton, Group Chief Sustainability Officer at Westfalia Fruit International.
The Importance of Accurate Yield Estimation
Precise yield forecasts enable orchard managers to make informed decisions regarding resource allocation, harvesting schedules, labour management, storage planning, and marketing strategies. Traditional methods, often reliant on manual counting, are labour-intensive, time-consuming, and prone to inaccuracies, especially in large-scale operations. Inaccurate estimations can lead to over or under-harvesting, resulting in increased waste and financial losses.
AI and ML Transforming Yield Estimation
The integration of AI and ML technologies has revolutionised yield estimation in orchards. These technologies analyse vast datasets, including images captured via drones, satellites, or ground-based sensors, to predict yields with remarkable accuracy. For instance, AI algorithms can identify and count fruits or flowers, assess their size and health, and monitor plant vigour, providing comprehensive insights into expected yields.
“AI alogorighms, driven by ML and fed directly to the edge (and then on to the cloud) can provide critical information that could make the difference between feeding our world or seeing it starve.” ~Dr Sarah-Jayne Gratton, Playing God with Artificial Technology, Camden Media (2024).
A notable example is the development of an AI system by researchers at the National Robotarium in Edinburgh. This system accurately estimates the number of flowers on fruit trees using standard smartphone photographs, aiding in predicting crop yields up to six months before harvest. Trials in Spanish peach orchards demonstrated a 90% accuracy rate, significantly outperforming manual counts, which typically have error rates between 30-50%.
Benefits to Growers and the Supply Chain
Implementing AI-driven yield estimation offers multiple advantages:
• Resource Optimisation: Accurate predictions allow for efficient use of water, fertilisers, and pesticides, reducing waste and environmental impact.
• Labour Management: Better forecasts enable precise planning of workforce requirements, ensuring adequate staffing during critical periods like harvesting.
• Market Planning: Reliable yield data supports strategic marketing and sales efforts, aligning supply with demand and potentially securing better pricing.
• Waste Reduction: By anticipating yields more accurately, growers can minimise overproduction and post-harvest losses, contributing to a more sustainable supply chain.
Case Studies and Real-World Applications
Several projects have showcased the effectiveness of AI in yield estimation:
• Agerpix System: This technology employs AI algorithms to identify and measure fruit, assess leafiness and vigour, and provide yield estimations. Pilots in top apple-producing regions analysed over 140 hectares with a precision of 90-95%, demonstrating the system’s efficacy in diverse conditions.
• Farmonaut’s Precision Farming Solutions: By leveraging AI-powered satellite imagery and data analytics, Farmonaut has enabled farmers to increase crop yields by up to 30% in certain regions. Their system offers real-time crop health monitoring, smart irrigation management, and predictive pest and disease management, all contributing to enhanced productivity and sustainability.
Challenges and Future Directions
Despite the promising results, challenges remain in the widespread adoption of AI for yield estimation:
• Data Requirements: AI models necessitate large volumes of high-quality data for training, which can be a barrier for some growers.
• Cost and Accessibility: The initial investment in AI technologies and the need for technical expertise may deter smaller operations.
• Adaptability: AI models must be adaptable to different crops, regions, and growing conditions to be universally applicable.
Ongoing research aims to address these challenges by developing more user-friendly and cost-effective solutions, ensuring that the benefits of AI-driven yield estimation can be realised across the agricultural sector.
Ultimately, the adoption of AI and ML in orchard yield estimation represents a significant advancement towards efficient and sustainable agriculture. By providing accurate, timely, and actionable insights, these technologies empower growers to optimise their operations, reduce waste, and enhance profitability, ultimately benefiting the entire supply chain.
