GREEN HOUSE RPI

ABSTRACT:-

The research paper has successfully implemented a comprehensive, complete package in the form of IoT based Smart Greenhouse employing a novel combination of Monitoring, Alerting and Automation and Disease Prediction using Deep Learning. The results developed, using both from internet and live testing, have high accuracy and employ memory efficiency. Future implementation of this system can be to include more number of sensors for improvement of data by including more parameters and installing multiple cameras at distinct points. This would generate a much wider database and would further aid in honing the results. However, there might be a possibility that its portability could be limited in that case. Further, with academic industry partnership this system can be made available to the farmers with small land holdings and encourage them to opt for greenhouses so that they don‟t bear the brunt of pests, droughts and floods and can minimize their crop loss.

As India is an agricultural country, it produces big agricultural merchandise. But the trouble raises with want of unseasonal meals merchandise during the year. For this availability of unseasonal agricultural merchandise like culmination and veggies during the yr the greenhouse systems are being designed. Greenhouse protection is a difficult activity for each proprietor and the labors, as a special and non-stop care want to be taken in preserving weather parameters like temperature, humidity, soil moisture, mild intensity, and carbon dioxide. This makes the greenhouse consequences in big loss and very low yield and productivity. As Embedded Systems and IOT are used to locating an answer for this trouble, IOT is the today’s and maximum usable records era used during the arena for Global cloud connectivity. Embedded gadget is ideal at control of sensors and controlling the actuators and devices. In this proposed prototype we’re combining the embedded structures with the IOT for computerized tracking and controlling of the greenhouse with none human interference. For this gadget we’re the use of Raspberry Pi3 as controller and sensors as inputs with threshold values for temperature, humidity, mild intensity, soil moisture and carbon dioxide. When threshold values are crossed then the controlling of the climatic parameters is carried out and facts is saved in ThingSpeak cloud era and records is despatched to the proprietor thru GSM. A Pi digital digicam is likewise located in prototype with a view to see the sphere of greenhouse while needed. So this consequences in computerized tracking and controlling with out human interference and consequences accelerated crop yield.

EXISTING SYSTEM:-

The current structures of the greenhouse farming aren’t that so powerful in retaining the farming loose from losses and exact crop yield. Those structures are simply having the sensors for informing the climatic situations in the greenhouse, however now no longer controlling them. Only the showing of the situations and sending SMS to the proprietor cellular via GSM is done. This system effects in human interface in controlling of climatic parameters. So we flow to advocate the system which controls them with out human interface.

PROPOSED SYSTEM:-

Greenhouses need maintenance, just like your plants. Greenhouses require regular maintenance and an ideal environment for plants to grow. Management of water leaks, humidity, lighting, ventilation, cooling and shading should not be neglected at any cost. To maintain a commercial greenhouse, growers need automatic monitoring instead of manual control. Sensors are key elements of greenhouse monitoring systems.As each conditional environment requires its own input, manufacturers must consider the requirements with the number of inputs available.

Temperature Sensors: Temperature monitoring systems collect temperature data through various sensors. Because temperature sensors are needed for many purposes, whether wired or wireless, it’s important for growers to determine the type of sensors they will be using. Wireless temperature sensors use a thermistor to accurately determine temperatures. The temperature measurement range is from 20°C to +55°C with 0.5°C tolerance.
Humidity Monitoring System: The Humidity Monitoring System increases productivity and prevents crop losses in the event of infection due to incorrect combination of humidity in the greenhouse. Moisture sensors help growers solve problems earlier and work in real-time mode. The estimated humidity range is 0% to 100% with a tolerance of ±3.5% RH.For growers, the handy humidity sensors are waterproof and dust proof and can be used both indoors and outdoors.
Co2 detection system: Co2 complements photosynthesis and plant production. As plants consume CO2, gas levels decrease, and a noticeable decrease could noticeably affect the bottom line. Co2 absorbs infrared light at a wavelength of 4.26 μm.If an infrared radiation segment is absorbed by a gas that contains CO2, this absorption can be calculated.In this way, most of the energy in the greenhouse comes from cooling and heating. Solar Panels Equipped with electricity to power swamp coolers, the color shading of the solar panels eliminates the need for blinds and keeps greenhouses cooler.
Soil Moisture Monitoring System: The Soil Moisture Sensor uses IoT to enable farmers to maximize yield, reduce disease and optimize resources. Depending on the root structure of each plant, multiple probes can be installed at different depths. Growers can monitor soil surface temperature and humidity using different types of IoT sensors.Data collected from IoT sensors can be pushed to the cloud for in-depth analysis, visualization and trending. An IoT-based sensor gives growers control and access to their greenhouse system anywhere with high-speed, real-time data updates. Even implementing IoT enabled sensors in a greenhouse saves costs, time and effort in growing crops and productivity.

IoT sensors empower ranchers to gather various pieces of information with phenomenal granularity. They give ongoing data on basic environment factors like temperature, moistness, light frequency and carbon dioxide all through the nursery. In our project we collect the data from IoT sensors like from DHT11 we collect humidity and temperature value then we connect soil moisture sensor to Arduino and collect moisture data from it, also we connect co2 and LDR sensors to and from that sensors we collect the data. We use these data to analyse and take action based on it. For that we have to integrate arduino and website part. For integration we use flask framework. Flask is a one type of framework which is available in python to integrate python with web(HTML, CSS). By using flask we integrate our project features on website which is useful to monitor greenhouse.

Our first module is to detect disease from leaf. So in this module we use neural network algorithm to identify and classify that disease. When user upload any leaf image on our user-frinedly website then we feed that image to trained neural network algorithm(CNN). A digital camera or similar device is used to take different types of pictures and then they are used to identify the affected area on the leaves, then different types of image processing techniques are applied to them, processing these pictures to obtain different ones and useful functions required for this purpose. The identification of plant leaf diseases is particularly necessary in order to predict both the quality and the quantity of the first segmentation step, which is mainly based on a smooth polygonal leaf model that is first created and then used to calculate the development of an energetic contour By combining the global shape descriptors of the polygonal model with local curvature features, the leaves on the back are classified in data sets. The first segmentation step based on the graphical cutting approach is carried out first and then used for guidance. The development of leaf boundaries and implementation of classification algorithms to classify diseases and recommend fertilizers for affected leaves. The leaves are attacked by bacteria, fungi, viruses and other insects. The algorithm classifies the image of the sheet as normal or affected. like color, shape, textures. Then hyperplane was created with conditions to categorize the preprocessed leaves and also to implement a multiclass classifier to predict diseases in the leaf image with greater precision.

Our second module is to detect which crop is suitable for greenhouse environment. For that we collect the live data from IoT sensors which are connected to arduino. Then we feed to trained machine learning algorithms which predict some values like N,P,K and crop name. Precision agriculture facilitates in reduction of non-suitable crop that so will growth productivity, apart from the following benefits like efficacy in enter in addition as output and better selection making for farming. This approach offers answers like providing an advice machine through an ensemble model with majority preference strategies mistreatment random tree, SVM and Naive Bayes as learner to recommend suitable crop supported soil parameters with excessive precise accuracy and potency. The categorized image generated via way of means of those strategies includes floor reality statistical data and parameters of it vicinity unit weather, crop yield, state and district sensible crops to expect the yield of precise crop under particular weather condition. The layout consists of collection period of time records and constructing the prediction model so creating a user interface for giving inputs. On the begin data preprocessing is completed. Once the pre-processing is completed, the usage of ML. set of rules model is generated for prediction. The test records is given to the generated model for prediction. The model is examined towards random enter values supported accuracy rate and error created while predicting throughout testing. Until the error price is decreased and accuracy rate is improved this approach is repeated. In order to accumulate the input from user, web application is employed. The inputs accrued are given to the crop prediction model for predicting the appropriate crops.

MODULES:-

User Register and Login: User have to register and login to check which elements (NPK) are suitable for soil and which crop will grow based on soil moisture and humidity. Also farmer can manage temperature, water level, moisture from home
NPK Prediction: We use crop dataset which is open-source to trained models (Adaboost) and predict NPK values.
Crop Prediction: Feeding NPK values to trained machine learning models such as Random forest, SVM, Decision tree to predict crop. Algorithm also recommend fertilizers to the user based on crop prediction.
Crop yield prediction: Based on user district ,season area and crop algorithm will predict yield and also recommend crop.
Monitoring greenhouse: Displaying realtime values by fetching values from IoT sensors. Also plotting a graph based on realtime values to monitor.
Disease Detection: Plant disease image feature extraction done with the help of CNN model to classify which type of disease it is.
CNN model generation: Plant disease categories will be added and trained with convolution neural network.

ADVANTAGES :-

Manipulation of Growing Seasons
Production volume increases more than 10-12 times than normal production
Round the year production of most desired
Disease and pest attack is minimum.
Suitable for rearing /hardening of issues culture plants

HARDWARE AND SOFTWARE REQUIREMENTS

HARDWARE:

Processor: Intel Core i3 or more.
Hard disk: 250 GB or more.
DHT11 Sensor
Soil moisture Sensor
Mq2 Gas Sensor
LDR Sensor
Raspberry pi

SOFTWARE:

Operating System : Windows 10, 7, 8.
Python
Anaconda.
Spyder, Jupyter notebook, Flask.
MYSQL.