Precision agriculture is a farming management concept based on observing, measuring and responding to inter- and intra-field variability in crops. In this paper, we focus on responding to intra-field variability in potato crops and analyse variable rate applications (VRAs). We made an overview of potential VRAs in potato crop management in The Netherlands. We identified 13 potential VRAs in potato, ranging from soil tillage to planting to crop care to selective harvest. We ranked them on availability of ‘proof of concept’ and on-farm test results. For five VRAs, we found test results allowing to make a cost-benefit assessment. These five VRAs were as follows: planting, soil herbicide weed control, N side dress, late blight control and haulm killing. They use one of two types of spatial data: soil maps or biomass index maps. Data on costs and savings of the VRAs showed that the investments in VRAs will pay off under practical conditions in The Netherlands. Savings on pesticide use and N-fertilizer use with the VRAs were on average about 25%, which benefits the environment too. We foresee a slow but gradual adoption of VRAs in potato production. More VRAs will become available given ongoing R&D. The perspectives of VRAs in potatoes are discussed.

Robotic plant-specific spraying can reduce herbicide usage in agriculture while minimizing labor costs and maximizing yield. Weed detection is a crucial step in automated weeding. Currently, weed detection algorithms are always evaluated at the image level, using conventional image metrics. However, these metrics do not consider the full pipeline connecting image acquisition to the site-specific operation of the spraying nozzles, which is vital for an accurate evaluation of the system. Therefore, we propose a novel application-specific image-evaluation method, which analyses the weed detections on the plant level and in the light of the spraying decision made by the robot. In this paper, a spraying robot is evaluated on three levels: (1) On image-level, using conventional image metrics, (2) on application-level, using our novel application-specific image-evaluation method, and (3) on field level, in which the weed-detection algorithm is implemented on an autonomous spraying robot and tested in the field. On image level, our detection system achieved a recall of 57% and a precision of 84%, which is a lower performance than detection systems reported in literature. However, integrated on an autonomous volunteer-potato sprayer-system we outperformed the state-of-the-art, effectively controlling 96% of the weeds while terminating only 3% of the crops. Using the application-level evaluation, an accurate indication of the field performance of the weed-detection algorithm prior to the field test was given and the type of errors produced by the spraying system was correctly predicted.

No abstract

In the last few years there is an increase in technology developments in agriculture using machine vision and deep learning to recognize specific plants or animals. A shared infrastructure to exchange image datasets and to support the workflow of image processing with neural networks could fasten up the developments of new vision-based applications in agriculture. This requires some form of standardization and architecture principles. The use case of plant specific weeding with robots is used to define an initial architecture for this infrastructure and to describe an initial set of preferred metadata for standardizing the exchange of image datasets and deep learning algorithms.

No abstract