Sunteți pe pagina 1din 10

Plant Growth Classification and Disease Detection

March 14, 2019

Plant Growth Classification March 14, 2019 1 / 10


Overview

1 Works Done Till Date


MatLab
OpenCV
CNN
Dataset

Plant Growth Classification March 14, 2019 2 / 10


MatLab

MATLAB is a multi-paradigm numerical computing environment.


MATLAB allows matrix manipulations, plotting of functions and data,
implementation of algorithms, creation of user interfaces, and
interfacing with programs written in other languages, including C,
C++, C, Java, Fortran and Python.

Plant Growth Classification March 14, 2019 3 / 10


MatLab

Enhance Contrast of Image


RGB image is converted to HSL/LAB colour space.
Image segmentation is done using K-Means Clustering.
Since the image has 3 different colours, 3 different clusters are created
Diseased leaves could be classified using SVM classifier.

Plant Growth Classification March 14, 2019 4 / 10


OpenCV

OpenCV (Open source computer vision) is a library of programming


functions mainly aimed at real-time computer vision.
OpenCV-Python makes use of Numpy, which is a highly optimized
library for numerical operations with a MATLAB-style syntax.

Plant Growth Classification March 14, 2019 5 / 10


OpenCV

The idea was to classify images using KNN


Since the inputs are images, it has to be converted to a feature vector
for classification using KNN or K-Means clustering.
So the idea was to mask different disease regions with the most
dominant colour. The the region can be represented as a single colour
region.
It is similar to constructing a dataset like IRIS.
Since the image contain different diseases it is not feasible to convert
to set of values.

Plant Growth Classification March 14, 2019 6 / 10


CNN

In deep learning, a convolutional neural network (CNN, or ConvNet)


is a class of deep neural networks, most commonly applied to
analyzing visual imagery.
CNNs use a variation of multilayer perceptrons and requires minimal
preprocessing.

Plant Growth Classification March 14, 2019 7 / 10


CNN

Collecting the Dataset


A large dataset is required so that the model can learn from them by
identifying certain relations.
Building the CNN
Convolution extract features from the input image.
Convolution preserves the spatial relationship between pixels by
learning image features using small squares of input data.
Pooling reduces the dimensionality of each feature map but retains the
most important information.
After pooling comes flattening. Here the matrix is converted into a
linear array so that to input it into the nodes of our neural network.

Plant Growth Classification March 14, 2019 8 / 10


CNN

Data Augmentation
If we have only a limited number of images for training the network,
minor alterations can be made to the dataset to get more data.
Minor changes include flips, rotations and translations. The neural
network willthink of these altered images s new images.
Data augmentation is a way we can reduce overfitting on models,
where we increase the amount of training data using information only
in our training data.

Plant Growth Classification March 14, 2019 9 / 10


Dataset

Dataset of tomato leaves categorized into different sets based on


diseases.
This dataset was previously used in Kaggle Machine Learning
competition therefore it is a valid dataset.

Plant Growth Classification March 14, 2019 10 / 10

S-ar putea să vă placă și