Professional Documents
Culture Documents
Artificial Intelligence Assisted Potato Disease Detection
Artificial Intelligence Assisted Potato Disease Detection
net/publication/353572164
Artificial Intelligence Assisted Early Blight and Late Blight Potato Disease
Detection Using Convolutional Neural Networks
CITATIONS READS
0 137
2 authors:
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Natnael Tilahun on 30 July 2021.
Abstract
Developing countries like Ethiopia have resources suitable for the production of varieties
of crops. Potato is the fourth major grown crop in the world, after rice, wheat, and maize.
In Ethiopia, one of the crops produced and consumed in mass is potatoes. Nonetheless,
the yield per unit area of potatoes is very low compared to other countries. There are a
plethora of reasons and one of them is potato disease. The major disease, which affects
the major potato production area is late blight, according to researchers on the field
estimated losses range from 6.5 to 67.7% depending on the accessibility of varieties. As
plant pathologists mentioned not to take early late blight disease management would
destroy the whole farm within a few days. For decades many researchers have
experimented on plant disease detection and classification using computer vision via
different approaches and algorithms. Most researchers used traditional machine learning
algorithms that require a handcrafted feature extraction to classify and detect a given
image as per its classes. The contribution of this work is twofold, using deep learning for
potato disease detection and developing an AI-based android application prototype. An
Image dataset that is labeled with three classes as ‘Healthy’, ‘Early blight’, and ‘Late
blight’ is used as a benchmark. The pre-trained models of deep learning, MobileNet, and
EfficentNet have shown 98% prediction performance. Finally, the model was built
integrated with the android application and tested with unseen data.
[15]
Ethiop. J. Crop Sci. Vol 8 No.2, 2020
crops. In the ruler area, about 68 the economy and basic needs,
percent of the total production was especially in Ethiopia where the
used for household consumption, and mainstream of the population is a
only 20 percent sold in the market. farmer.
Furthermore, due to its nutrition,
Ethiopia is considered a strategic Plant disease can cause social,
commodity for ensuring food security economic, and ecological losses (Ufaq
(Lupescu and Zimmerman, 2015) Khan and Ashish Oberoi, 2019).
Appropriate plant disease detection at
Potato is one of the crops, which is a an early stage prevents the devastation
faster-growing crop not only in of the farm and the risk of food
Ethiopia but also in other sub-Saharan security of human beings. Some
countries. The total area of land diseases are challenging to detect crop
covered with potatoes has increased disease through naked eye
dramatically from 500,000 Mg to observation. One of the challenges is
about 3,700,000 Mg in 10 years. The the similarity of the diseases of the
growth of potato production allows the crop, in this case, to be sure of the
country to achieve the food security exact disease of the crops requires a
program. Nonetheless, the productivity laboratory test by a plant pathologist.
of the crop is low with an average The problem challenged farmers and
yield of around 12.3 Mg/ha, the added effort on experts. Therefore, to
amount is very low compared to the overcome this problem researchers
attainable yields of 50 Mg/ha using studied computer vision with machine
improved farm management with learning and it achieved promising
improved varieties. The research results. Currently, the state of the art
identified Bacteria Wilt and Late technology uses deep learning which
Blight as the major potato production is a subset of machine learning and
constraints and the number one disease achieved excellent results in plant
that the farmers manage (Tafesse et disease detection using different parts
al., 2018). of the plant.
Tadesse Demissie (2019) reported that Ethiopia has a good climate and soil
Ethiopian farmers‟ know-how about cultivation for many different types of
potato Late Blight disease that only a crops including potato, and it grows in
few farmers have information about different regions with various agro-
potato‟s Late Blight disease (Tadesse economic conditions. Moreover,
Demissie, 2019). This implies the need potato is one of the important cash
for technological solutions to support crops, which gives ready cash to
farmers and individuals who are farmers. It is also known as „complete
investing in the agriculture sector. food‟ as it contains carbohydrates,
Furthermore, the problem has a great proteins, vitamin B, vitamin C, and
impact on developing countries minerals like phosphorus, calcium, and
because agriculture is the backbone of iron required for body growth. It is the
[16]
Ethiop. J. Crop Sci. Vol 8 No.2, 2020
richest source of starch, unlike other Kenya. The disease is highly studied
crops, it produces more food per unit and it is the most destructive one. It
area in a short time (Tadesse was responsible for a loss of $5 billion
Demissie, 2019). annually and became a threat to food
security (Tadesse Demissie, 2019).
The production of potato crops is Unless necessary management of the
affected by biotic and abiotic factors. disease is taken, it could destroy 100%
Among those, the major factor is the of the farm. Therefore, developing
biotic factor, that caused by countries like Ethiopia need early
microorganisms like bacteria, fungus, detection of Late Blight disease, to
and viruses. Many potato diseases save people‟s life and economy.
affect the growth and production of the
crop, to mention some of the diseases Artificial Intelligence (AI): refers to
are Early Blight, Late Blight, Septoria the simulation of human intelligence
Leaf Spot, and Virus. Besides, several in computers designed to think and
factors affect potato yield, and that‟s imitate humans‟ behavior. The term
called abiotic factors. To mention can also be applied to any system that
some, lack of well-performing shows characteristics linked to a
cultivars, poor fruit setting due to human mind, such as learning and
heavy rains, and excessive-high problem-solving. The concept of AI is
temperature, and pets. based on the idea that human
intelligence should be described in
Late Blight disease is the major such a way that a computer can easily
disease that affects potato farms of imitate it and perform tasks, from the
Ethiopian farmers compared to other easiest to the most complicated ones.
potato diseases. Consequently, the Machine learning, deep learning, and
yield per unit of potato is low relative CNN is a subset of AI. Figure 1
to countries like Rwanda, Egypt, and presents the AI and its subclasses.
[17]
Ethiop. J. Crop Sci. Vol 8 No.2, 2020
Rule Classical
Answer
Data Programming
Data
Machine Learning Rule
Answer
[21]
Ethiop. J. Crop Sci. Vol 8 No.2, 2020
Figure 4: Training and testing accuracy Figure 3: Training and testing loss
[23]
Ethiop. J. Crop Sci. Vol 8 No.2, 2020
Figure 5 presents the accuracy of the correctly, and finally in the Healthy
model per each class. It was class out of 23 images 22 images
interpreted as, in Early Blight class out detected accurately by the model. The
of 150 images used for testing 148 accuracy of the model for each class is
images were detected correctly, and in summarized in Table 1.
Late Blight class out of 150 images
used for testing 148 were detected
[24]
Ethiop. J. Crop Sci. Vol 8 No.2, 2020
Artificial Intelligence Assisted Early Blight and Late Blight Potato Disease
Detection Application
SelectModel
<<include>>
ViewDiseaseType
Users
<<include>>
SelectDevice
[25]
Ethiop. J. Crop Sci. Vol 8 No.2, 2020
[26]
Ethiop. J. Crop Sci. Vol 8 No.2, 2020
[27]
Ethiop. J. Crop Sci. Vol 8 No.2, 2020
Tadesse Demissie. 2019. Integrated Potato Sardogan, M., Tunc, A., and Ozen, Y.
( Solanum Tuberosum L .) Late 2018. Plant Leaf Disease Detection
Blight ( Phytophthora Integrated and Classification Based on CNN
Potato ( Solanum Tuberosum L .) with LVQ Algorithm. UBMK 2018 -
Late Blight ( Phytophthora Infestans 3rd International Conference on
) Disease Management in Ethiopia. Computer Science and Engineering,
(December). 382-385.
doi:10.11648/j.ajbio.20190706.16 doi:10.1109/UBMK.2018.8566635
Durmus, H., Gunes, E. O., and Kirci, M. Singh, D., Jain, N., Jain, P., Kayal, P.,
2017. Disease detection on the leaves Kumawat, S., and Batra, N. 2020.
of the tomato plants by using deep PlantDoc: A dataset for visual plant
learning. 2017 6th International disease detection. ACM International
Conference on Agro-Geoinformatics, Conference Proceeding Series, 249-
Agro-Geoinformatics 2017. 253. doi:10.1145/3371158.3371196
doi:10.1109/Agro- Tafesse, S., Damtew., E., Mierlo, B. v.,
Geoinformatics.2017.8047016 Lie, R., Lemaga, B., and Sharma, K.
FAO. 2018. Ethiopia: small family farms 2018. knowledge and practices of.
country factsheet. From NJAS - Wageningen Journal of Life
www.fao.org/family-farming/data- Sciences(September 2017), 0-1.
sources/dataportrait/farm-size/en doi:10.1016/j.njas.2018.03.004
Ferentinos, K. P. 2018. Deep learning Tan, C., Sun, F., Kong, T., Zhang, W.,
models for plant disease detection Yang, C., and Liu, C. 2018. A survey
and diagnosis. Computers and on deep transfer learning. Lecture
Electronics in Agriculture, 145, 311- Notes in Computer Science
318. (including subseries Lecture Notes in
doi:10.1016/j.compag.2018.01.009 Artificial Intelligence and Lecture
Ufaq Khan and Ashish Oberoi. 2019. Notes in Bioinformatics), 11141
Plant Disease Detection Techniques: LNCS, 270-279. doi:10.1007/978-3-
A Review. International Journal of 030-01424-7_27
Computer Science and Mobile Tyagi Vipin. 2018. Understanding Digital
Computing, 8(4), 59-68. From Image Processing.
www.ijcsmc.com doi:10.1201/9781315123905
Oppenheim, D., Shani, G., and Erlich, O. Yan-Tak Ng. (2020). Deep AI. From Deep
2019. Using deep learning for image- Learning Course With Andrew NG:
based potato tuber disease detection. https://www.youtube.com/watch?v=
Phytopathology, 109(6), 1083-1087. yofjFQddwHE
doi:10.1094/PHYTO-08-18-0288-R Lupescu, M. and Zimmerman, J. 2015.
Prasanna Mohanty, S. 2020. PlantVillage- Potatoes and Potato Strategic
Dataset. From GitHub: Analysis and Intervention Plan. 1-11.
https://github.com/spMohanty/Plant
Village-Dataset
[28]