Professional Documents
Culture Documents
Submitted by
VAIBHAV BHOSALE
AKNOWLDGEMENT
To deal with the stated challenges, a deep learning-based approach has been
introduced to recognize and classify the paddy crop stresses. The proposed
methodology consists of two main stages, namely, image dataset preparation and
deep learning-based classification.
In this project we are trying to classify the type pf crop from the image.
Index
2.5 Objectives of
proposed system
4.2 Conclusion
INTRODUCTION:
To contribute to the development of nations, knowledge of agriculture sectors is crucial.
Agriculture is a one-of-a-kind source of wealth that develops farmers. For a strong country,
the development of farming is a necessity and a need in the global market. The world’s
population is growing at an exponential rate, necessitating massive food production in the
next 50 years.
The paddy crop cultivation is an integral part of Indian agricultural economy ranking first in
the area of cultivation with 43.92 million hectares with the production of 111.50 million
tons (Anonymous, 2018). Although the overall numbers are impressive, the rice production
has come under increased pressure in Asia due to population growth and changing socio-
economic factors. The major paddy producing belts in Asia regions achieves only 40% of
total production efficiency due to damage caused by drought, diseases, and pests. A more
integrated approach involving optimum crop improvement and resource management
practices such as nutrient management, irrigation regime, and other agronomic
management factors for paddy crop with the existing farmers' practices is a potential option
for minimizing the crop yield gap.
Today, we have Agriculture data set with us for better understanding and we’ll be focused
on how to think about these projects, rather than their implementation as many of us
getting trouble in initiating and doing the ending of projects.
LIMITATIONS:
Some of the limitations that currently exist with these simulators are that: there is no single
program or set of model parameters that have been identified for all global regions and crops,
there's limited precision to quantify crop responses to micronutrient stress, and there is a
limited amount of validation data ...
To deal with the stated challenges, a deep learning-based approach has been
introduced to recognize and classify the paddy crop stresses. The proposed
methodology consists of two main stages, namely, image dataset preparation and
deep learning-based classification.
In this project we are trying to classify the type pf crop from the image.
Methodology adopted, aims at predicting the crop yield of a particular area by
providing the required dataset from that area. It involves the following steps.
• Collection of the dataset for agricultural crop mapping and monitoring.
• Performing the pre-processing on the satellite image for extraction of data features
for further analysis.
• Classifying the type of crop using various approach.
• Applying the machine learning for accurate results.
Dataset Description:
Training data:
Dataset (Crop Images) contain 40+ images of each Agriculture crop(Maize, Wheat,
jute, rice and sugarcane)
Dataset (kag2) contains 159+ augmented images of Crop Images of each
class. Augmentation contain Horizontal flip, roatation, horizontal shift,
vertical shift. Agriculture crop images | Kaggle
Testing data:
https://www.kaggle.com/aman2000jaiswal/testssss
Tools/Technology: Python
Platform: Kaggle
Packages: numpy, pandas, sklearn, matplotlib, tensorflow, keras
Modelling: Deep learning (vgg model)
The proposed work has employed the pre-trained VGG-16 CNN model for the
classification. The VGG model, or VGGNet, that supports 16 layers is also referred to
as VGG16, which is a convolutional neural network model. In VGG architecture, all
the convolutional layers use filters of the size of 3 x 3 with stride =1 and same
padding, and all the max-pooling layers have a filter size of 2 x 2 with stride = 2. With
each set of a convolutional layer, the number of filters doubles and with each
pooling layer, the width and height of the image reduces by half.
1. Input: The VGGNet takes in an image input size of 224×224. For the ImageNet
competition, the creators of the model cropped out the center 224×224 patch in
each image to keep the input size of the image consistent.
2. Convolutional Layers: VGG’s convolutional layers leverage a minimal receptive
field, i.e., 3×3, the smallest possible size that still captures up/down and
left/right. Moreover, there are also 1×1 convolution filters acting as a linear
transformation of the input. This is followed by a ReLU unit, which is a huge
innovation from AlexNet that reduces training time. ReLU stands for rectified
linear unit activation function; it is a piecewise linear function that will output
the input if positive; otherwise, the output is zero. The convolution stride is fixed
at 1 pixel to keep the spatial resolution preserved after convolution (stride is the
number of pixel shifts over the input matrix).
3. Hidden Layers: All the hidden layers in the VGG network use ReLU. VGG does
not usually leverage Local Response Normalization (LRN) as it increases memory
consumption and training time. Moreover, it makes no improvements to overall
accuracy.
4. Fully-Connected Layers: The VGGNet has three fully connected layers. Out of
the three layers, the first two have 4096 channels each, and the third has 1000
channels, 1 for each class.
Step1: Load the data (Set up the working directories, initialize the images, resize, and
perform test-train splits)
Step2: Configure the model (Data augmentation, build the model, set up the
callbacks and other hyper-parameters)
Step3: Train the model (Train and monitor the progress)
Step4: Evaluation and prediction (Get a classification report and predict class labels
of new unseen images)
5. Results:
The complete system is designed using Python. The output of the used algorithm
shows an accuracy of 0.94.
Operating environment
TOOL JUPYTER
CHAPTER 4:
LIMITATIONS:
The obtained result for the crop yield prediction using SMO classifier gives less accuracy
when compared to naïve Bayes, multilayer perceptron and Bayesian network.
Previously yield is predicted on the bases of the farmers prior experience but now weather
conditions may change drastically so they cannot guess the yield.
Companies need a large or high-quality small dataset to have accurate classifications
Noise (e.g., emojis, slang, or punctuation marks) can reduce accuracy
Costs are higher compared to traditional, rule-based methods.
DRAWBACKS
The Difference in Growing conditions
Certain locations and their climates are more favorable for monoculture, meaning a certain
kind of crop. Other crops, other than that specific type of crop, cannot grow well in that
specific type of temperature and soil conditions.
Obligatory Crop Diversification
For crop rotation to work, one has to plant different crops every time. Nonetheless, it does
not allow a farmer to specialize in a single type of crop. The farmer is not able to produce a
single crop on a large scale over a long period of time because of the damage it will do to
the soil.
Improper Implementation Can Cause Much More Harm Than Good
Improper implementation of this technique causes much more harm than good. If one lacks
the technical know-how of crop rotation, there is no need to experiment because there can
be nutrient buildup that will take a longer time to correct. One has to have the skills to know
what crops can be planted after the other and in which season for the process to be
successful.