The 14th International Conference on Precision Agriculture (ICPA) presented by the International Society of Precision Agriculture (ISPA) was held in Montreal, Quebec, Canada during the last week of June. Some information about the conference, including how the association is seeking a definitive definition of precision agriculture, was already published in another article. I will highlight some of the topics presented by keynote speakers, as well as the research covered in the oral presentations.
Chris Paterson, Bayer CropScience Digital Farming, was the first keynote speaker. His presentation focused on some trends that are driving adoption of precision agriculture and farm data technologies on North American farms. Some companies, including Bayer, are facing a big transformation from input providers to service providers in order to take advantage of the new scenario agriculture will face in the near future. Collaboration with other companies is a key factor to enable this. Partners from a broad spectrum of technologies, including farm data management platforms, equipment sensors, weather and imagery sensors, robotics, artificial intelligence, and wireless mobile connectivity, are joining efforts to speed up development. Some potentially disruptive technologies to watch for include the ones that will enable crop protection on an individual plant basis, with potential input savings of more than 90% and greater environmental appeal.
Dr. Yoshua Bengio was the second keynote speaker. He is scientific director of Montreal Institute for Learning Algorithms (MILA) and one of the pioneers in deep learning. In his presentation, “From Data to Decisions with Artificial Intelligence,” he highlighted some of the recent achievements of machine learning techniques and how they are affecting important areas, including medicine and agriculture. Most of the questions from the audience were related to the importance of data in the successful implementation of these technologies. Most methods cannot deal with unstructured and non-standardized datasets, which are very common in precision agriculture. The minimum number of observations and the restriction imposed by data privacy policies are also a concern.
The applications of machine learning and big data in precision agriculture are relatively new, although much has been speculated by big companies and numerous startups. Scientific publications that have addressed the topic have been dominated by the adaptation of methods developed for other areas. Nevertheless, at least 10% of oral presentations at ICPA were grouped into the big data, data mining, and deep learning category. One of the reasons for the fast development of deep learning is the availability of free and open-source datasets, libraries, and coded implementations of most papers published. These are still hard to find when related to precision agriculture applications. Some initiatives have been developed, mostly towards gathering data from different sources to form a standard dataset to be accessible for all contributors and to be open to everyone at some point in the future. This is an important step to initiating the scientific development of data intensive models.
MORE BY RODRIGO TREVISAN
The oral presentations included many other research topics: Site-Specific Nutrient, Lime and Seed Management; In-Season Nitrogen Management; Geospatial Data; Decision Support Systems; On Farm Experimentation with Site-Specific Technologies; Proximal and Remote Sensing of Soil and Crop, including Phenotyping; Applications of Unmanned Aerial Systems; Robotics, Guidance and Automation; Wireless Sensor Networks; Smart Weather for Precision Agriculture; Drainage Optimization and Variable Rate Irrigation; Precision Crop Protection; Precision Horticulture; Precision Dairy and Livestock Management; Farm Animals Health and Welfare Monitoring; Site-Specific Pasture Management; Education and Outreach in Precision Agriculture; Precision Agriculture and Global Food Security; Profitability and Success Stories in Precision Agriculture; Small Holders and Precision Agriculture and Land Improvement and Conservation Practices. More information about the research presented at the conference and future conferences can be found on the ICPA page of ISPA website.
This was the third time I attended this conference and the transformation and evolution of scientific research during the last several years has been incredible. Most of the technologies developed over the last 20 years were hardware based, and its adoption is still low. We are experiencing a change to service-based solutions, both in research and commercial products, which can be adopted at a much faster rate, as soon as they are ready to meet farmer’s expectations.