Site Loader
Rock Street, San Francisco

Oil-palm (Elaeis guineensis) trees are an
important economic crop which are the source of palm oil, a widely used
vegetable oil in the world. Its other major uses include furniture, plywood,
paper etc. Crop diseases are the major source of food insecurity and famine at
a global scale. Not only that, farmers also have to face disastrous
consequences as their livelihood is dependent on healthy crops.  Like most of the crops oil-palm trees are also
prone to diseases.  Common ones include
Ganoderma butt rot and Oil palm wilt caused by fungus (Turner, 1981).  These devastating diseases cause direct loss
of stand and reduction in yield. Trees with the symptoms of these diseases die
within 1-2 years.  Symptom of these
diseases include pale yellow foliage of leaves. With the progress of disease
the palm show retarded growth with leaves turning brown and spear leaves
remaining unopened, ultimately causing the death of palm (Hushiarian, Yusof, & Dutse, 2013). The
infected tree has to be quarantined and removed to prevent the spread of
disease (Singh, 1991).

Early detection of the diseased trees is vital
to manage the disease effectively and to prevent the spread of disease. Manual
inspection of the palms in order to monitor their health is very time consuming
and expensive. Remote sensing provides time and cost efficient solutions for
precision agriculture.

We Will Write a Custom Essay Specifically
For You For Only $13.90/page!

order now

A lot of research has been done on oil-palm
management using satellite based remote sensing data for identifying and counting
oil-palm trees in a farm (Shafri, Hamdan, & Saripan, 2011; Srestasathiern
& Rakwatin, 2014) and
identifying diseased regions in image (Santoso, Gunawan, Jatmiko, Darmosarkoro, &
Minasny, 2011; Zulhaidi, Shafri, & Hamdan, 2009). These
task require high accuracy and precision and it is more challenging using
satellite based data due to cloud cover being one reason amongst other and also
access of high resolution imagery is costly. Aerial images are the good
alternative when it comes to high spatial resolution.

The use of Aerial imagery for precision
agriculture has increased due to the availability of high resolution imagery
through UAVs and airborne source (Berni, Zarco-Tejada, Suárez, González-Dugo, &
Fereres, 2009; Rokhmana, 2015). Technological strategies using machine vision and artificial
intelligence are being investigated to achieve intelligent farming using high
resolution aerial images. For managing oil-palm plantation,
machine learning has been used for identifying and counting oil-palm trees in a
farm using aerial imagery (Malek, Bazi, Alajlan, AlHichri, & Melgani, 2014;
Miserque Castillo, Laverde Diaz, & Rueda Guzmán, 2016). Deep
learning is a subset of machine learning which deal with the set of algorithm inspired
by the working of brain, known as artificial neural networks. In remote
sensing, Deep learning has been used in many applications like building
detection  (Vakalopoulou, Karantzalos, Komodakis, & Paragios,
road detection (Mnih & Hinton, 2010), vehicle
detection (Chen, Xiang, Liu, & Pan, 2014) , image
classification (Li, Fu, Yu, Gong, et al., 2016) and scene
classification (Hu, Xia, Hu, & Zhang, 2015) using
remote sensing imagery.

Convolutional Neural Networks (CNN) are a
category of Neural Networks that have proven very effective in areas such as
image recognition (Krizhevsky, Sutskever, & Hinton, 2012; LeCun,
Kavukcuoglu, & Farabet, 2010). In
precision agriculture CNN has been used for disease management in detection of Ceratocystis wilt in Eucalyptus crops
from aerial images (Souza et al., 2015).

In case of oil-palms CNN have been used to count
the trees in remote sensing images (Cheang, Cheang, & Tay, 2017; Li, Fu, Yu, &
Cracknell, 2016) however
no effort has been applied for detecting diseased oil-palm trees using CNN. This
study is an attempt to evaluate the performance of convolutional neural
networks in detecting diseased oil-palm trees in Aerial images of the
plantations in Ecuador and Indonesia.

Oil-palm plants are usually cultivated in
large farm which can spread up to many hectors. To monitor the health of oil-palm
plants, farmers have to traverse through large areas for the inspection. This
inspection could be very time consuming and demands a labor force which could
be costly. There is a need of a system which could automatically detect diseased
plants in there early stages of disease with a coordinate level accuracy, so
that they can treated in time, thus decreasing the financial loss to the
farmer. The lack of previous studies in detecting diseased oil-palm plants
through aerial imagery encouraged to evaluate state-of-the-art technique of
image recognition using deep learning for detecting diseased oil-palm plants in
aerial imagery.

Deep learning is the area of machine learning
research which deals with the algorithms motivated by the working of biological
brain known as artificial neural networks, bringing machine learning closer to
its objective of artificial intelligence. Figure 1.1 illustrates the
relationship between different areas of artificial Intelligence using a Venn
diagram with an example.

Figure 1.1: A Venn diagram
illustrating domains of AI. (Ian Goodfellow and Yoshua Bengio and Aaron Courville, 2016)


In deep learning computational models are
composed of multiple processing layers also known as hidden layers that are
capable of learning a representation of data with multiple levels of abstractions
(LeCun, Bengio, & Hinton, 2015). The
hierarchical architecture of these models gave the word “deep” to such class of
machine learning algorithm. These algorithm extract less complex features in
the earlier stages and combine those feature with high level of abstraction in
later stages to learn a representation with minimum human intervention.

Convolutional neural networks (explained more in
detail in section 3.3) are the
descendants of artificial neural networks that can learn complex feature from
large amount of image data and have proven to be state-of-the-art in many
vision tasks. Because of their successful implementation in many image
recognition task in the domain of remote sensing they are promising technique
for detecting diseased oil-palm plants in high resolution aerial images.  

Post Author: admin


I'm Dora!

Would you like to get a custom essay? How about receiving a customized one?

Check it out