Page 81 - IJEEE-2023-Vol19-ISSUE-1
P. 81
Abdulla & Marhoon | 77
motor, and then the value changes between 100-300, opening and closing the water pump according to the reading
meaning the soil is wet. But if the sensor reads another value, alert message. Fig. 14 IV shows the comparison of our
it is moving from the soil, and its reading is wrong. system with previous related work.
Fig. 12: Read sensors data. Fig.14: Comparing the proposed system with a group of
previous methods.
ab c
V. CONCLUSIONS
Fig. 13: Irrigation system where (a): sensor read dry soil,
(b): motor on and soil is wet, (c) represent error reading for At the end of this work, one can conclude that there is
an attempt to create a smart farm utilizing the power of Deep
the sensor. Learning and IoT techniques. Both are integrated via
developing a mobile application using android studio tools.
IV. DISCUSSION The proposed system enables the farmer to manage his farm
without needing residency inside the farm. The resultant
The monitoring system designed using deep learning smart farm collects the environmental information that
and IoT is proven effective at work. In terms of deep affects crop status utilizing the IoT assets. analyze this
learning, we have discovered its effectiveness in classifying information to issue timely alerts to the farmers through the
tomato diseases with high accuracy of 97%, even with field mobile application. This alert enables the farmer to make
images that lack training. Thus, the farmer can rely on it to suitable decisions to protect his crop from disasters. Thus,
determine the plant's type of disease with high accuracy. In increasing the quality and quantity of production while
terms of monitoring using sensors, the system has proven its reducing the losses, improving the human labor used, and
effectiveness in reading sensor values and giving alerts when reducing the challenges facing the farms at present, including
tested. Thus, the monitoring system is like a farmer's eye on the increase in temperatures, the change in the level of
his farm. He can monitor it remotely by alerting him if the rainfall, the lack of water, and Plant diseases and the
read values of temperature, humidity, soil moisture, water difficulty of identifying them and others.
quality, and co2 exceed the permissible limit for the plant.
The automatic irrigation system was the complementary part And by using deep learning, the farmer can easily identify
of this system. For the farmer to have the ability to control the type of disease that the plant has. Our system designed to
and monitor his farm fully, the system proved its ability to classify tomato diseases exceeded the accuracy test 97% by
irrigate crops remotely through the farmer’s control of using 15,918 images from PlantVillage and google images
after augmenting it to solve unbalanced and A few field
images in google. And thus, it can be relied upon in
classifying and taking the appropriate action to prevent the
spread of the disease without the need to call agricultural
experts, thus saving more time and money.
By using the sensors for measuring the environmental
conditions surrounding the plant, the farmer can monitor his
farm remotely without needing to visit it in the field, and he
can irrigate his farm remotely using the automatic irrigation
system. The work of the application is not limited to
monitoring conditions only. It also can send alerts to the
farmer by reading the values of the sensors and comparing
