IEEE Consumer Electronics Magazine - March/April 2023 - 113

FIGURE 3. Accuracy comparison of different DL
algorithms for fertilizer recommendation.
chemical cost reductions, possible production
increases, and environmental preservation.
The applications above are geared at the
FIGURE 2. Workflow diagram of the DL-based
fertilizer recommendation.
namely the service and network layer of the smart
farming architecture. WSN data will be maintained
in a database accessible online.9
Digital image processing techniques may generate
various vegetation indices based on
images, which can lead to conclusions, either for
each shot individually, about the crops' status.10
Usually, crop absorption and scattering in different
electromagnetic spectrum bands are combined
mathematically into vegetation indices.11
Vegetation indices can be used to determine relevant
crop features based on the vegetation's
critical biological and physical properties.
Application Layer: The farmer may connect
with the mobile apps to remotely manage the
harvesting process due to the suggested precision
intelligent farming system architecture.
Sowing and irrigation operations and plant protection
and cultivation methods12 are examples
of uses in the agricultural area. Fertilizer recommendation,
spraying, weed detector mapping,
irrigation for field and alert system are some of
the applications that may be implemented.
Especially, the fertilizer recommendation
application aims to optimize nutrient utilization
by determining the quantity of fertilizer to apply
depending on the plant's health. This use has
various advantages, including fertilizer and
March/April 2023
effective field and crop management, with the
following goals in mind.
Boost production.
High product quality.
Make better use of chemicals in agriculture.
Keep pesticide levels under control.
Control water usage and underground water
levels to preserve the land.
IOT AND AI-EMPOWERED FERTILIZER
RECOMMENDATION APPLICATION
In the application layer of smart farming system,
the fertilizer recommendation service is
one of the major applications that focus in this
work. Among the various agricultural issues, suggesting
the necessary amount of fertilizer helps
farmers reduce their money on extra fertilizer
and corresponding workforce. Generally, including
different crops, three essential fertilizers are
in practice, namely, urea (N), single super phosphate,
and unit of potash muriate of potash
(MOP). The right amount of fertilizer for land
can be identified based on NPK values. Thus, the
recommendation is based on soil nutrient value
extracted by the sensors and stored in a cloud
environment.
In the conventional farming method, the combination
of fertilizers and each fertilizer's level
was decided randomly or depended on the farmer's
financial status, which may lead to soil erosion
and wastage of resources. Thus, in this
work, we applied ML and DL algorithms for
113
IEEE Consumer Electronics Magazine - March/April 2023

Table of Contents for the Digital Edition of IEEE Consumer Electronics Magazine - March/April 2023
