Intelligent Plant Growth Monitoring

and Management System

Anna Varghese Feba Ann Kurian Nivya Johny Mrs. Smitha Cyriac
Dept.of Electronics and Dept.of Electronics and Dept.of Electronics and HOD
Communications Engineering Communications Engineering Communications Engineering ECE Department
Viswajyothi College of Viswajyothi College of Viswajyothi College of Viswajyothi College of
Engineering & Technology Engineering & Technology Engineering & Technology Engineering & Technology
Vazhakulam, Muvattupuzha, Vazhakulam, Muvattupuzha, Vazhakulam, Muvattupuzha, Vazhakulam, Muvattupuzha,
Kerala, India Kerala, India Kerala, India Kerala, India
[email protected] [email protected] [email protected] [email protected]

Abstract - Process-based models are capable of them. The system gives a timely alert to the user’s
simulating plant biophysical processes considering Android phone by gathering and analyzing the data
the effects of soil, weather, management, and about the changing soil moisture condition, weather, and
genetics. However, there is a limitation to these temperature with the help of various sensors. All the
models mainly due to availability, uncertainty, and parameters like humidity, temperature, soil, moisture,
spatial variation of model parameters. To solve these light intensity, and pH, are tracked by the system, and
problems, Intelligent plant growth monitoring and this information is uploaded to the database. This system
management system is developed. This paper also helps to monitor the plant’s condition from any part
automates plant monitoring and management system of the world like the user’s workplace. This intelligent
in the Raspberry Pi platform. The main purpose of plant growth monitoring system will continuously
this system is to provide comfort to the people by monitor the condition of the plant and alert the user to
reducing manual work and to improve the overall make the changes that require immediate action for the
performance of any system without manual plants.
interaction. The important parameters for the quality
and productivity of plant growth are soil and air
II. LITERATURE REVIEW
temperature, humidity, sunlight, soil moisture, and
pH. Information to the user about the plant's health A. Visual Classification of Lettuce Growth Stage
and growth may be provided to the user by based on Morphological Attributes using
continuously monitoring and recording these various Unsupervised Machine Learning models [1]
parameters. It provides a better understanding of
how each parameter affects the growth of plants. All This study specifically solves the predicament of
the sensors (Temperature, moisture, humidity, pH...) identifying the developmental growth of plants
used in this project interface with a Raspberry Pi from seed leaves to amend the techniques of plant
controller. And the information about the growth science and cultivation management. With a view
and health of the plant can be directly monitored and to this, the paper shows coupled color-based
controlled by the owner through his/her smartphone. superpixels and multifold watershed transformation
Keywords - Moisture, pH, Temperature, in segmenting the lettuce image from the
Humidity, Light Intensity, Raspberry Pi. background. To fathom it out, a comparative
analysis of three unsupervised machine learning
algorithms. These were done by modeling each
I. INTRODUCTION algorithm from the features extracted from
morphological computations of the lettuce images
Intelligent plant growth monitoring and management
raised in a smart aquaponics setup. Each of the
system will help to ease the most tedious job of
models was optimized to increase cross and hold-
gardening for plant lovers. The system brings more
out validations.
convenience and comfort to people’s lives by taking care
of their plant’s growth. The proposed system is designed
and implemented using a low-cost credit-card-sized
Raspberry Pi, which is monitored through a remote
monitoring sensor. This helps to solve plants thrive by
tracking the environmental conditions and monitoring
B. Machine learning approaches for growth current days, farm monitoring is still contingent on
monitoring using multi-temporal and multi-variety the instinctive human decisions showing low
remotely sensed data [2] accuracy due to the visual limitations of the human
eye and lacking knowledge and specialization.
The main goal of this paper is to build a machine Unfolding the issue, Machine vision can be a
learning model based on the data derived from the considerate field that involves making a machine
sensor, with a combination of agro-meteorological ‘‘see”.
data. This methodology will also consider the
temporal variation of the input data set acquired in
III. PROPOSED WORK
multiple development phases along the crop season.
The dataset used for modeling was built from two Intelligent management of agriculture systems,
kinds of sources. The first source is sugar cane aiming at optimizing profitability, productivity, and
sucrose data from ripening sugar cane fields, sustainability, comprises a set of technologies including
obtained for a different plot across three harvest sensors, information systems, informed management,
seasons, from 2014-to 2016. The second source is etc. The main purpose of this system is to provide
remotely sensed data. The performance of the built comfort to the people by reducing manual work and
model and previous model will be evaluated improving the overall performance of the system without
according to the error between estimated and manual interaction. An intelligent plant growth
observed values. monitoring system will help to ease the tedious job of
gardening and it will help to solve plants thrive by
C. Classification of Chili Plant Growth using Deep
tracking the environmental conditions and monitoring
Learning [3]
them. The system also gives a timely alert to the user by
The paper discussed the potential deep learning gathering and analyzing the data.
algorithm in classifying chili plant growth images
from a small dataset. The experiment is performed IV. PROJECT ELEMENTS
on 256 chili plant images under various conditions,
where the images were captured using the camera. This chapter deals with the various hardware and
The data is divided into two models training and software elements used in the project.
testing. The results showed that both models were
A. Hardware Units
able to detect the correct age of the chili plant with
96% accuracy. a) Raspberry Pi 4 Model B - 4GB RAM
Raspberry Pi 4 Model B is the latest product in the
D. A Smart Agriculture System [4] popular Raspberry Pi range of computers. This
product’s key features include a high-performance
The system has an easy-to-upgrade bank of
64-bit quad-core processor, dual-display support at
inference rules to control the agriculture resolutions up to 4K via a pair of micro-HDMI ports,
environment. This system mainly looks at inputs, hardware video decode at up to 4Kp60, up to 8GB of
such as temperature, humidity, and pH. The system RAM, dual-band 2.4/5.0 GHz wireless LAN,
is ubiquitous as it enables distant access. This Bluetooth 5.0, Gigabit Ethernet, USB 3.0, and PoE
system is an addition to current-state-of-art capability.
Internet-of-things. b) Logitech C270 HD Webcam
The Logitech C270 HD Webcam has features such as
E. An automated detection and classification of citrus
HD 720p video calling and HD video recording, 2.4
plant diseases using image processing techniques:
Giga Hertz Intel Core2 Duo, 2 GB RAM, 200 MB
A review [5] hard drive space with video capture upto 1280x720
pixels and crisp 3 MP photos technology.
A thorough literature review of image
c) Capacitive Soil Moisture Sensor
preprocessing, segmentation, and feature extraction
using machine vision affirms to lessen the threats Soil moisture can be measured using a variety of
and requisition of the detection and classification different techniques: gravimetric, nuclear,
electromagnetic, tensiometric, and hygrometric,
systems. It also authenticates the efficiency and
among others. The technique explored here uses a
high accuracy compared to other methods. But still,
gravimetric technique to calibrate a capacitive-type
the automated vision-based detection and electromagnetic soil moisture sensor. Capacitive soil
classification methods are in their infancy stage. In moisture sensors exploit the dielectric contrast
between water and soil.
 d) DHT11 Digital Humidity & Temperature Sensor
The DHT11 is a basic, ultra low-cost digital
temperature and humidity sensor. It uses a capacitive
humidity sensor and a thermistor to measure the
surrounding air and spits out a digital signal on the
data pin.
e) Light Dependent Resistor (LDR)
Light Dependent Resistors (LDR) are also called
photoresistors. They are made of high-resistance Fig 2. Schematic Diagram for PCB
semiconductor material. When light hits the device,
the photons give electrons energy. This makes them
VI. WORKING
jump into the conductive band and thereby conduct
electricity. To monitor plant growth, the process-based model
f) pH Sensor Module is widely used to stimulate plants' biophysical processes
This PH Sensor Module is compatible with Arduino. using agrometeorological data, soil properties, and plant
The pH stands for the power of hydrogen, which is a management. The limitation of these models is mainly
measurement of the hydrogen ion concentration in the due to the availability, uncertainty, and spatial variation
body. This is used in water quality testing and of model parameters. To solve these problem empirical
Aquaculture. The total pH scale ranges from 1 to 14, models built using a machine learning algorithm is
with 7 considered to be neutral. A pH less than 7 is
developed to monitor the growth and health of the plant.
said to be acidic and solutions with a pH greater than
The performance of the built model is evaluated with a
7 are basic or alkaline.
comparison of the statistical model and machine learning
B. Software Unit model. By analyzing the above studies, here we are
planning to design an Intelligent plant monitoring and
a) Google Colaboratory
management system with the help of a Raspberry Pi
Colaboratory, or “Colab” for short, is a product from
controller. In this system, various parameters like
Google Research. Colab allows anybody to write and
temperature, humidity, light, and pH. are controlled with
execute arbitrary python code through the browser
and is especially well suited to machine learning, data the help of a sensor. The growth and health of the plant
analysis, and education. More technically, Colab is a can be directly controlled and monitored by the owner
hosted Jupyter notebook service that requires no from any part of the world via the internet. Our block
setup to use, while providing access free of charge to diagram consists of a monitoring area. The plant to be
computing resources including GPUs. monitored is placed in this area. And there is a plant
health monitoring system. It consists of sensors and
V. BLOCK DIAGRAM & SCHEMATIC actuators. It will sense various parameters like
temperature, humidity, pH, and light. . A camera is
provided and it can move 360 degrees and it will capture
the images of the monitoring plant. The data from the
monitoring area and health monitoring system will be
given to the machine learning model. Here Raspberry Pi
4 is used. A machine learning model is trained data. Here
the data given to the machine learning model is saved
into two folders test and train. Then this tested data is
compared against the trained data. If the error rate is near
0.1, then the model is assumed to be trained otherwise an
anomaly is detected. The data is then given to another
device via the internet. So the user can monitor the
plant's condition from any part of the world.

Fig 1. Block Diagram

 VII. METHODOLOGY TABLE III
Moisture Results
Create a dataset that contains the images of
tomatoes. For this, we take clear and unclear images
of tomatoes from different angles with different
variations and kept them in a common folder. But
this is difficult in our case, so we take an open-
source tomatoes image. Thus the primary base
model is done with this open-source dataset.
These images are of different sizes so first, we resize
all the images into a standard size of 640*640. This TABLE IV
Light Intensity Results
is because coding is done using the Yolo algorithm.
This provides graphical performance and algorithm
performance of the system.
Draw a bounty box for detecting objects or classes.
label IMG is the library for labeling. For this, we
need to browse the path of the folder where the
image is stored. Simultaneously images are loaded
one by one. By drawing the bounty box, the image is
saved by the same name as the image. TABLE V
pH Results
The bounty box data is saved. XML format. Once
these steps are completed, the process is stopped.

VIII. RESULTS

This chapter deals with the results obtained after

running 10 test images through the algorithm.

TABLE I
Temperature Results
IX. CONCLUSIONS

Intelligent plant growth monitoring and

management system will help to ease the most tedious
job of gardening for plant lovers. The system brings
more convenience and comfort to people’s lives by
taking care of their plant’s growth. The proposed system
is designed and implemented using a low-cost credit-
card-sized Raspberry Pi, which is monitored through a
TABLE II
Humidity Results remote monitoring sensor. This helps to solve plants'
thrive by tracking the environmental conditions and
monitoring them. The system gives a timely alert to the
user’s Android phone by gathering and analyzing the
data about the changing soil moisture condition, weather,
and temperature with the help of various sensors. All the
parameters like humidity, temperature, soil, moisture,
light intensity, and pH, are tracked by the system, and
this information is uploaded to the database. This system
also helps to monitor the plant’s condition from any part
of the world like the user’s workplace. This intelligent
plant growth monitoring system will continuously
monitor the condition of the plant and alert the user to
make the changes that require immediate action for the
plants.
 X. REFERENCES

[1] Jonnel Alejandrino"Visual Classification of Lettuce

GrowthStage based on Morphological Attributes

using Unsupervised Machine Learning models"2020

IEEE REGION 10 CONFERENCE

(TENCON)Osaka, Japan, November 16-19, 2020
[2] "Machine learning approaches for crop growth

monitoring using multi-temporal and multi-variety

remotely sensed data"YU ZHAO; DIEGO DELLA

JUSTINA HITACHI, Ltd.; University of Campinas
[3] A. D. A. Aldabbagh, C. Hairu, M. Hanafi

"Classification of Chili Plant Growth using

DeepLearning"2020 IEEE 10th International

Conference on System Engineering and Technology

(ICSET), 9 November 2020, Shah Alam, Malaysia

Tomen:A Plant monitoring and smart gardening
[4] system using IoT March 2018 International Journal

of Pure and Applied mathematics volume 119

(No. 7)

Z. Iqbal, M. A. Khan, M. Sharif, J. H. Shah, M. H.
[5] ur Rehman, and K. Javed, “An automated detection
and classification of citrus plant diseases using
image processing techniques: A review,” Comput.
Electron. Agric., vol. 153, no. August, pp. 12–32,
2018.