Abstract:
Agriculture sector is a major economic force in
Sri Lanka, making a significant contribution to the national
economy. However, it is observed that land and human
resources available in Sri Lanka for agriculture becomes scarce.
On the other hand, the demand for high quality agricultural
products is on the rise. Therefore, we need to embrace cost
efficient and effective technologies that would facilitate the
automation of large scale plantations to obtain an increased
yield per given area. Providing water, nutrients, pesticides and
other supplies is one of the most resource consuming tasks in a
large scale plantation. Existing solutions to achieve this objective
are only capable of providing supplies from time-to-time
without knowing the exact requirement of the plant. This may
result in a shortage or excess amount of supplies that will
adversely affect the health of plants. In response, the proposed
method is capable of identifying the exact status and health of
plants to customize the supplies accordingly and ensure the
healthy growth of plants. When it comes to this project, almost
the total management of the plant is done by the system from
the beginning to end. Initially the system decides the age of the
plant and learn about the required conditions. An array of
sensors and cameras connected through Narrowband Internet
of Things (NB-IoT) will collect data that will be processed using
machine learning techniques. Initially image processing is
supposed to do the object detection and to perform the leaf
counting task it is intended to consider the capability of deep
convolution neural network. Automatic detection of these type
of issues in plants is supposed to achieve by a software solution
for automatic detection and classification of plant leaf diseases.
This would facilitate the understanding of plant growth and any
development of pests or weeds. Water, nutrients and pesticide
will be supplied according to the exact requirement of plants. In
addition, the proposed system will alert if immediate attention
is required in case of an emergency and also report on the
growth of plants and usage of each type of supplies to facilitate
a thorough insight of the plantation. Because, when it comes to
a traditional farm, farmers have to pay their attention to each
and every plant separately to identify factors like diseases, insect
attacks or any other infections. Therefore, farmer’s special
attention have to be taken in such kind of situations. It is
envisaged that the proposed system will play a pivotal role is
automating large scale plantations that would in return benefit
the agriculture sector In Sri Lanka.