Download as pptx, pdf, or txt
Download as pptx, pdf, or txt
You are on page 1of 20

Technology: Python

Domain: DL & ML

E-AGRI KIT- AGRICULTURE AID USING


DEEP LEARNING
CONTENT

• Problem statement • Project flow diagram


• Abstract • Hardware and software
• Introduction requirements
• Existing work • Architecture
• Disadvantages • Algorithms
• Proposed work • Modules
• Advantages • System designing
PROBLEM STATEMENT:

• Agriculture plays a crucial role in the global economy and is vital for food security. However, farmers face
numerous challenges in optimizing their agricultural practices, managing resources effectively, and
maximizing crop yields. Traditional farming methods often lack access to real-time data, advanced
analytics, and technology-driven solutions, resulting in inefficient resource allocation and suboptimal
decision-making.
ABSTRACT

Our paper presents an agriculture aid application which is developed and designed to help the farmers. As
Agriculture was the key development in the rise of sedentary human civilization, whereby farming of
domesticated species created food surpluses that enabled people to live in cities. And it will be difficult
for a person to identify every crop just by looking at the leaves and along with that selling the crop for a
suitable and affordable prices is difficult to a farmer. Where to overcome this all problems we are
creating an application where we are developing an algorithm that detects the type of crop by giving the
leaf input image and also creating a platform where an investor can invest into a crop by funding into a
crop which is given by a farmer and where farmer can sell their crops to buyers with the rates that which
are suitable with the crop.

KEYWORDS: Agriculture, Deep learning, Investment, Farmers, selling.


INTRODUCTION
According to a study by the Associated Chambers of Commerce and Industry of India, annual crops losses
due to pests and diseases amount to Rs.50,000 crore ($500 billion), which is tantamount to a country where
at least 200 million go to bed hungry every night. Agriculture being a vital sector has a majority of the rural
population in developing countries relying on it.

The sector is faced by major challenges like unprecedented pest attack and unforeseen weather conditions
affecting their produce leading to major loss of food and effort. Technology plays a vital role in uplifting the
livelihoods of the rural populace which can be done by using a simple agro-android application system.

Plant diseases can affect vast produce of crops posing a major menace to food security as well as leading to
major losses to farmers. An extensive review of existing research was conducted by us on this domain and
in an effort to help farmers overcome this problem, we have designed an android application,
EXISTING SYSTEM
Identifying the different crops is difficult to a normal person who has no knowledge
on agriculture and the plants. While if a person wants to make the investment into the
agriculture part and if don’t have any knowledge about agriculture it will be difficult
without having the guidance from a farmer and selling the crop is only in the hands
of third party people where farmers do not have the permissions to sell their crop
which could be a loss to farmers if those were sold with improper rates.
DISADVANTAGES

• No knowledge in identifying crops.

• Farmers may get losses.

• It take more time

• Human involvement more

• There is no data management


PROPOSED SYSTEM

To overcome the problems from the existing system we are creating an application where we are
developing an algorithm that detects the type of crop by giving the leaf input image and also
creating a platform where an investor can invest into a crop by funding into a crop which is given by
a farmer and where farmer can provide the details regarding the crop that which includes farmer
name, crop name, crop yield, location, year, investment, profit also can view money for investing in
a crop and then farmer can provide the agreement to the investor. And where farmer can sell their
crops to buyers with the rates that which are suitable with the crop. If buyer is fine with the selling
details provided by farmer buyer can contact farmer to buy the crop and buyer make payment after
purchasing the crop.
PROJECT FLOW
ADVANTAGES

• Easy prediction of crops.


• Accurate classification
• Easy way to sell the crop
• Advantageous to investors

• Online communication reduce the time and human involvement

• Here data stored in proper way for that we are using MYSQL Database
SYSTEM SPECIFICATIONS

SOFTWARE SPECIFICATIONS:

Technology : Python, Application.

Libraries : Pandas, Numpy, TensorFlow, OS.

Version : Python 3.6+

Server side scripts : HTML, CSS, JS

Frame works : Flask

IDE : Pycharm

HARDWARE SPECIFICATIONS:

RAM : 8GB, 64 bit os.

Processor : I3/Intel processor


ARCHITECTURE
ALGORITHM
• 3.Training: (CNN with MobileNet)

• We are using the pre-processed training dataset to train our model using CNN algorithm.

• CNN algorithm consists of 4 layers: Input layer, Convolution Layer, pooling layer, Flatten layer and dense layer.

• In input layer we consider images as input.

• In Convolution layer, we convert image into matric format. Here matrix size is 1024 X 1024 (rows X columns).

• In the pooling layer the numerical values will be stored. To change the numerical data to binary data, we use machine
learning algorithm named Softmax (supervised learning algorithm). In Softmax layer we will convert the numerical data to
binary.

• In flatten layer and dense the classes of total dataset (17 types) is stored which will be in the binary data format.

• We use fit_generator method for saving the data in the form of .h5. Here model is a format for storing the binary data.
ALGORITHM
• CNN with Resnet50:

• We are using the pre-processed training dataset to train our model using CNN algorithm with Resnet50 model.

• CNN algorithm consists of 4 layers: Input layer, Convolution Layer, pooling layer, Flatten layer and dense layer.

• In input layer we consider images as input.

• In Convolution layer, we convert image into matric format. Here matrix size is 1024 X 1024 (rows X columns).

• In the pooling layer the numerical values will be stored. To change the numerical data to binary data, we use
machine learning algorithm named Softmax (supervised learning algorithm). In Softmax layer we will convert
the numerical data to binary.

• In flatten layer and dense the classes of total dataset (17 types) is stored which will be in the binary data format.

• We use fit method for saving the data in the form of .h5. Here model is a format for storing the binary data.
ALGORITHM
• Random forest regression:

• Irrespective of the algorithm we select the training is the same for every algorithm.

• Given a dataset we split the data into two parts training and testing, the reason behind doing this is to test our model/algorithm
performance just like the exams for a student the testing is also exam for the model.

• We can split data into anything we want but it is just good practice to split the data such that the training has more data than the
testing data, we generally split the data into 70% training and 30% testing.

• And for training and testing there are two variables X and Y in each of them, the X is the features that we use to predict the Y
target and same for the testing also.

• Then we call the .fit ( ) method on any given algorithm which takes two parameters i.e., X and Y for calculating the math and
after that when we call the .predict ( ) giving our testing X as parameter and checking it with the accuracy score giving the
testing Y and predicted X as the two parameters will get our accuracy score and same steps for the classification report and the
confusion matrix, these are just checking for how good our model performed on a given dataset.
MODULES
1. Farmers:

Registration : At first farmers will register in to the website to provide the details for investors, the details include crop name,
farmer name, address, crop type, in which year crop is harvested, investment, profits on investment, username[email], password,
mobile num.

Login: Once after registration they will login and can perform the following.

 Upload the data for funding’s like crop name, crop yield, location, farmer information, money for investing.
 Will enter details for selling a crop, the details include framer name, ph no, location, crop name, crop quantity, crop price.
Agreement: Once ok with the investor farmer need to provide the agreement form to the investor.
2.Investors:
• Register: Investors will first register into the website with their details.

• Login: Once after registering, they can login with their credentials.

• View crop information: Once after login, investor can view the funding data provided by farmer like crop name, crop yield,
location, farmer information, money for investing. If investor is fine with everything they will send request to farmer for
investing.
MODULES

 If farmer accepts the request, investor can invest.

 Investor can view the updates from the farmer

3. Buyers:

Register: Buyers who wants to buy the crop will register at first with their details

Login: Once after registering, they can login with their credentials.

View selling details: Once after login, they will view the selling details from the farmers and will contact to farmer.

Make payment: after purchasing the crop buyer need to pay the amount to the farmer.

SYSTEM:
1. Create Dataset:
•Here we are searching the dataset with the help of internet. For example, we consider the kaggle website mainly for the datasets. If data is
unavailable in the website we our self create our data with a technique called data augmentation.

MODULES
• Data Augmentation:

• Data augmentation in data analysis are techniques used to increase the amount of data by adding slightly modified copies of already
existing data or newly created synthetic data from existing data. It acts as a regularize and helps reduce over fitting when training a
machine learning model. It is closely related to oversampling in data analysis.

• The size of overall dataset is 13.74 GB. From which we have are splitting the data into (Training_images, Testing_images), where the
Training_images dataset size is 11.7 GB and Testing_images size is 2.04 GB.

• 2.Data Pre-Processing:

• •Data pre-processing is a data mining technique which is used to transform the raw data in a useful and efficient format.

• (a). Missing Data:

• •This situation arises when some data is missing in the data. It can be handled in various ways.

• (b). Noisy Data:

• •Noisy data is a meaningless data that can’t be interpreted by machines. It can be generated due to faulty data collection, data entry
errors etc.
MODULES
• Prediction:

• It loads the model that which was created from the training part, choose the images from the system.
And then the image is changed into array using img_to_array method.

• For changing binary values we perform matrix division operation. After converting the binary
values, we load binary converted data to model for prediction.

• After predicting we will get 17[Brown Spot, Healthy, Hispa, Leaf Blast, Corn Gray Leaf Blight,
Corn Common Rust, Corn Northern Leaf Blight, Corn Healthy, Potato Early Blight, Potato Late
Blight, Potato Healthy, Diseased Cotton Leaf, Healthy Cotton, Tomato Bacterial Spot, Tomato Late
Blight, Tomato Yellow Leaf Curl Virus & Tomato Healthy] values. But finally, we classify only one
image at a time. For this we use argmax method to get the highest value from the 17 classes.


INPUT AND OUTPUT
Input: here input communication between the farmer, buyer and
investor.

Output: leaf image classification, crop yield prediction, fund requests,


agreement, crop selling and payment

You might also like