IOT and Deep Learning based Plant Monitoring System and Diseases Classification

Agriculture plays a vital role in the development of any country. Countries whose economic system greatly depends on the agriculture need to pay more intention on agriculture sector in order to produce more yields from cultivated land. Pakistan?s economy is directly linked with the production of qua

Project Title

IOT and Deep Learning based Plant Monitoring System and Diseases Classification

Project Area of Specialization

Internet of Things

Project Summary

Agriculture plays a vital role in the development of any country. Countries whose economic system greatly depends on the agriculture need to pay more intention on agriculture sector in order to produce more yields from cultivated land. Pakistan’s economy is directly linked with the production of quality crops. The population of Pakistan has drastically increased tremendously so the most challenging task is to double the food production in order to meet basic requirement of our society.

In Agriculture fields, the environmental parameters also play a constructive or destructive role in maintaining required food quality and productivity. Heavy rain falls and high intensity ultraviolet sunlight during the summer can damage the food productivity. Our farmers use old and traditional methods which is another reason of low yield and poor quality grains. There is no any perfect mechanism for collection and monitoring of data for different environmental parameter. But IOT can play an important role in making our agriculture system very smart. We can collect all the environmental data through sensors and then we can send all data to cloud for analysis. Such useful data can be remotely accessed at real time from cloud into the MATLAB Thing Speak platform to visualize the collected data in graphical form for continuously monitoring.

Using deep learning inception v3 CNN Tanfers learning, we can classify diseases as well. Diseases may be early blight or late blight.

Project Objectives

• To reduces the Human interaction in agriculture farming because manual methods are less efficient and very costly.

• To automate the agriculture sector for better productivity and quality.

• It will facilitate the farmers by providing the insecticides free environment.

• Gathering the important environment’s parameters that are important for the health of the plants and predict the necessary precautionary measure in the big farms.

Project Implementation Method

PROJECT IMPLEMENTION METHOD

1. Nodemcu ESP8266

2.Temperature Sensor

3.Soil moisture sensor

4.Air pressure

5.LDR Sensor(Light dependent resistors)

6.Rainfall Sensor

1.Nodemcu ESP8266

NodeMcu ESP8266 WiFi module is a low cost and economical IOT controller which is widely used in embedded IOT applications. It simply works like as Arduino but due to the excellent features of Wi-Fi connectivity and a combination of digital pins which we can communicate with multiple devices. 

CPU: ESP8266

Storage: 4MBytes

Why is NodeMCU used?

NodeMCU is an open source platform based on ESP8266 which can connect objects and let data transfer using the Wi-Fi protocol. In addition, by providing some of the most important features of microcontrollers.

2.Temperature Sensor

It is basically digital in nature that helps in measuring the temperaturelevel from environment. In agriculture field, every plant needs some optimum temperature to grow well. This sensor helps in monitoring the temperature sensing from environment..

How does temperature sensor work?

The basic principle of working of the temperature sensors is the voltage across the diode terminals. If the voltage increases, the temperature also rises, followed by a voltage drop between the transistor terminals of base and emitter in a diode.

3.Soil moisture sensor

The device is equipped with a soil moisture sensor that measures soil moisture and the soil temperature. The sensor is located at the end of a cable with a waterproof housing and inside is protected with resin. The sensor has three steel elements which is in contact with the geological material.

working principle of soil moisture sensor?

The Soil Moisture Sensor uses capacitance to measure dielectric permittivity of the surrounding medium. In soil, dielectric permittivity is a function of the water content. The sensor creates a voltage proportional to the dielectric permittivity, and therefore the water content of the soil.

4.Air pressure

Air pressure sensor that is capable of measuring the pressure of air in the environment. The accuracy of BMP180 is ±0.12 hPa. We know that every plant needs some optimum air pressure to survive. We monitored the air pressure data through this sensor.

5.LDR Sensor(Light dependent resistors)

The working principle of this sensor is also depends resistance. When light intensity is high it lows down the voltage and when light intensity is too low it increases the voltage. It is a low cost sensor and its accuracy is depend upon resistance toward sunlight that is about 100?.

6.Rainfall Sensor

The purpose of using this sensor is to measure the contents of water drops. Most of the time in agriculture field there is heavy rainfall that decreases the food productivity. So, we monitor the rainfall data through this sensor and record that how much times the rainfall has occurred. It is low cost sensor which operate both in digital and analog mode.

Benefits of the Project

Benefits of the Project:

• In Pakistan there is no automation mechanism in agriculture sector. All the Agricultural productivity is based on manual labor force which is unable to meet the food demands of drastically growing population. The system will be cost effective and will help us in the improving production of crops and as well as minimize the possible health hazards to our farmers.

• Due to lack of information about the environment’s condition, our plant gets diseases which result in low production. Also the farmers’ naked eye observation is imprecise and contains inaccuracy. If we have the knowledge of that parameter we can take precautionary measures before the disaster.

• There is also possiblility that farmers only judge the health of the plants by the physical condition of leaves. Many diseases are not appear at the early stages so, it is needed a mechanism that can judge the health of plant and predict the upcoming disease of the plant. Unfortunately there is no mechanism has implemented so far.

Technical Details of Final Deliverable

IOT base Plant Monitoring System:

                                                            In IoT Part, All five sensors are connected with the NodeMcu. NodeMcu gets data from all sensors and sends it to the Thingspeak cloud via WiFi. Four Sensors are connected to a digital pin and one sensor connects with an analog pin due NodeMcu has only one analog pin. 

1. LDR, DHT22, BMP180, Rainfall, and relay connected with digital pin
2. Soil moisture sensor connected with analog pin
3. one action perform when the threshold value meet, the motor will ON.

Circuit Diagram:

ThingSpeakIOT Analytics Platform:

                                                               It is an easy and open-source MATLAB IoT Analytical Platform for the visualization of the wireless sensors data. This application is basically used for data monitoring and analysis. In ThingSpeak IoT Analytics Platform we make a GUI of all the sensors. Esp8266 and Arduino are easily compatible with the ThingSpeak IoT Analytics Platform. Using the API key, we can send sensor data to the ThingSpeak channel.

Deep Learning base Diseases Classification:

                                                            Early blight and late blight of potato are two widely distributed diseases. Both diseases are responsible for huge economic losses. Alternaria solani is the main causative fungus of early blight of potatoes. In contrast, Phytophthora infestans are the main causative agent of late blight in potatoes. So, this is the key difference between early blight and late blight of potatoes. Moreover, warm temperatures and high humidity favor early blight of potato while cool and moist weather favor late blight of potato. Both diseases produce brown spots on leaves and stems.

For the classification of diseases of the potato plant, we are using the Inception V3 CNN transfer learning Model classify the diseases. Early blight and late blight are two types of diseases. After identifying the diseases farmer can do the appropriate treatment. we used binary classification with sigmoid activation function and adam as optimizer. Data collect from Kaggle and trained the model. 

Final Deliverable of the Project

HW/SW integrated system

Core Industry

Agriculture

Other Industries

Core Technology

Internet of Things (IoT)

Other Technologies

Sustainable Development Goals

Decent Work and Economic Growth, Climate Action

Required Resources

If you need this project, please contact me on contact@adikhanofficial.com