Revolutionizing agriculture: The AI-driven future of farming

From precision livestock feeding to crop yield optimization: Exploring AI’s transformative impact on agriculture

Eight years ago, at the 2016 Gartner Symposium, I sat with an audience of CTOs, CEOs, and CIOs, all eager to glimpse the future as forecasted by the keynote speaker. The session opened with a captivating video, projecting the technological landscape of the next decade, with a special emphasis on artificial intelligence (AI) and its prospective advantages to the agricultural sector. I vividly recall Gartner’s predictions around advancements in farming technology – from precision seeding of fields to the intricate matching of weather trends with optimal planting times. 

Today, as we witness these predictions unfold into reality, the integration of AI within agriculture has transcended beyond the mere enhancement of machinery. It has ushered in a transformative era marked by efficiency, sustainability, and innovation. This evolution is not just about technological advancement but about securing a future where farming is more informed, adaptive, and capable of meeting the global demands of tomorrow.

 

As more growers realize the triple win that farm automation can represent—greater agricultural productivity and profits, improved farm safety, and advances toward environmental-sustainability goals—excitement about these technologies will spread.^[1]

 

In this blog post, we look at some customer stories and use cases for AI in agriculture: from revolutionizing seed genome analysis to optimizing livestock feed; from completely automating milking processes for increased frequency and cow wellbeing to employing visual AI to monitor aquaculture, AI’s role in agriculture is redefining our interaction with the environment. 

AI is a game changer for crop yields

Inari, a leading agricultural tech start-up, has made significant strides in revolutionizing crop yields with its AI-powered platform, earning it the title of 2023 Overall AgTech Company of the Year. Inari’s mission is to enhance the resilience of major crops such as wheat, soy, and corn against climate change impacts, aiming to increase yields by 20% and substantially reduce the use of land, water, and nitrogen. This ambitious goal challenges the traditional annual yield increase of about 1%.

The company’s journey involved overcoming obstacles related to the diversity of its team and the complexity of its data, particularly in genome editing through its proprietary SEEDesign™ technology. Its collaboration with SoftwareOne to develop a unified and automated machine-learning research platform marked a turning point. This platform, built on AWS, introduced end-to-end governance of data sets, unified workflow tooling, and democratized machine learning, significantly accelerating research and innovation.

Key benefits of the platform include a dramatic reduction in workflow execution times and a cost-effective "pay-as-you-go" model, thanks to technologies like Elastic Kubernetes Service and Databricks. The platform’s ability to auto-scale based on processing needs has led to considerable cost savings and efficiency improvements. For instance, experiments that once took 24 hours can now be completed in just 20 minutes.

This transformation has not only resulted in faster and more cost-efficient research but also improved the quality of science at Inari. The shared notebooks on Databricks facilitate collaboration and enhance visibility and consistency in experiments, allowing scientists to quickly validate ideas and secure intellectual property for Inari. 

Optimizing pig feed

Felleskjøpet Agri, a Norwegian agricultural cooperative, collaborated with Inmeta (a SoftwareOne company) to develop an innovative solution for optimizing pig feeding using data and AI. The project aimed to improve pig nutrition, health, and growth by finding the optimal mix of feed ingredients for each farm, considering factors such as feed availability, nutritional content, pig breed, growth rate, and feeding system constraints.

The solution leveraged mathematical optimization, Azure functions, Python tools, and Power Apps to create a cloud-based system that replaced the old manual, Excel-based approach. The solution also involved extensive consultations with feed advisors and farmers, ensuring the tool was user-friendly, practical, and customized for each farm.

The project resulted in significant benefits for Felleskjøpet and its customers, such as reduced feed costs, enhanced pig health and growth, increased efficiency and scalability, streamlined data management, and compliance with Norway’s animal welfare standards.  

Innovating and enhancing milking efficiency through computer vision

The dairy industry is experiencing a significant transformation thanks to the integration of computer vision and neural networks into the milking process, leading to streamlined operations and substantial financial benefits.

In the United States, where the average dairy farm houses approximately 300 cows, optimizing milking frequency is critical for maximizing profits, given that the industry generates around USD 150 billion annually across more than 36,000 farms. The advent of Automatic Milking Systems (AMS) since 1990 has allowed for economic scaling, increasing the average herd size per farm and necessitating more frequent milking to support the doubled milk production per cow over the last three decades.

A pivotal advancement came with the development of AlexNet, a convolutional neural network that significantly enhanced computer vision capabilities. BouMatic, a leading dairy equipment manufacturer, capitalized on this technology through a partnership with Inmeta to improve their milking systems, particularly focusing on teat detection during milking. The implementation of Computer Vision and Neural Networks led to a model that excelled in detection speed, localization accuracy, robustness, multi-detection, and pattern recognition.

The model processes 3D images to adapt to various cow types and lighting conditions, achieving a 99% prediction accuracy for teat positioning at speeds up to 31 frames per second. Such precision reduces attachment misses and shortens the time spent in the milking booth, significantly enhancing cow comfort and milk quality. The integration of machine learning and edge computing has further optimized the process, allowing for rapid, high-volume data processing.

The financial implications of these technological advancements are profound and by facilitating more efficient and less stressful milking processes, farms can not only increase their milk yield but also improve the overall health and productivity of their herds.  

Leveraging visual intelligence for enhanced agricultural monitoring

Visual intelligence monitoring represents a cutting-edge application of AI in agriculture, transforming the way operations such as aquaculture or livestock rearing are managed. A notable example is the use of AI combined with remote sensing by companies to accurately count and monitor shrimp farms through satellite imagery.

This innovative approach not only showcases the potential of visual intelligence in optimizing farm operations and increasing yields but also opens the door to a multitude of applications across the agricultural sector. AI and remote sensing are also making strides in areas such as livestock management such as counting of cows over wide areas in Australia and in viticulture, where forecasting the yield and quality of grapes is becoming more precise.