www.ijcrt.

org © 2022 IJCRT | Volume 10, Issue 10 October 2022 | ISSN: 2320-2882

IMAGE CLASSIFICATION ANDROID

APPLICATION
1
Mr.Pratik Jadhav, 2Mr.Ashish Kambale, 3Mr.Sairaj Matkar
1
Student, 2Student, 3Student
1Department of Information Technology,
1Finolex Academy of Management and Technology, Ratnagiri, India

Abstract: Image classification is used to close the visual perception gap between computers and humans, enabling machines to
recognise images in the same manner that people do.
Although the traditional system that is now in place has been used extensively in practical situations, there are still certain issues
with practical procedures, such as poor classification accuracy, poor adaptive capacity, and unsatisfying results. Therefore, deep
learning has been suggested for picture classification in order to address these issues and improve accuracy.
This essay examines the fundamental ideas around the use of deep learning for image classification, as well as the advantages of
deep learning for image classification and the numerous applications that employ deep learning for image classification.

Index Terms - python, Tensorflow Library, Convolutional Neural Network, Artificial Neural Network.
I. INTRODUCTION

Neural networks were first proposed in 1944 by Warren McCullough and Walter Pitts, two students from the University of Chicago
the researchers moved to MIT in 1952. Inventor The first neural computer, Dr. Robert Hecht-Nielsen, identified a neural network
in the form of - "... a computer system consisting of a number of simple and highly connected treatments elements process
information according to their dynamic state response to external inputs. Neural networks are used to machine learning, in which
the system learns to perform tasks by training set analysis. Thousands / millions interconnected processing nodes are used in NN.
Most these buttons are "feed-forward", which means only data move in one direction. Individual node receiving data multiple nodes
and send process data to other nodes. During training, initial node weights are assigned to value, when the training data is passed
to the NN, the weight is continuously adjusted until it gives the correct result. Deep learning is more advanced in a few years, many
technologies GIFTS like, GOOGLE, MICROSOFT, FACEBOOK, BAIDU is interested in deep learning. Basically, Deep Learning
is a subfield of machine learning that deals with the algorithm allows the software to train itself to perform image recognition, by
displaying a multi-layer neural network for large amounts of data, inspired by the function of The brain is called an artificial neural
network. Deep learning is good for identifying patterns in images. It uses the same neurons neat approach to various problems like
support Vector machine, linear classifier, regression, Bayesian, Decision trees, grouping and association rules. One an example of
deep learning is essentially how Facebook can automatically sort photos, recognize faces, and make recommendations Which
friends to tag.

Image classification is the process of assigning the pixels in a digital image into classes of interest. The objective of the image
classification is to identify the unique features of the image. In order to classify a set of data into different classes or categories, the
relation between the data and the classes into which they are classified must be well known. The image
classification is used in different areas like; eg. in satellite remote sensing the features are measurements made by sensing in
different wavelengths of the electromagnetic spectrum-visible/infrared/microwave/texture features etc.

II. LITERATURE REVIEW

Neural Networks are basically mathematical models to solve an optimization problem. They are made up of neurons, the basic
unit of computation of a neural network. He also called it Artificial Neural Network (ANN). There are different types of ANN -
Depending on the function of neurons in human brain and network, artificial neural network or ANN perform tasks in a similar
way. Most artificial neural networks will have some similarities with their biological counterparts that are more complex and very
efficient in their expected tasks, such as. segmentation classification.

IJCRT2210371 International Journal of Creative Research Thoughts (IJCRT) www.ijcrt.org d200

www.ijcrt.org © 2022 IJCRT | Volume 10, Issue 10 October 2022 | ISSN: 2320-2882
Convolutional Neural Network (CNN) is one of the most frequently used deep learning method used for image classification.
CNN learns directly from image data so there is no need of manual feature extraction. It has mainly three advantages such as
parameter sharing, equivalent representation and sparse interactions.
A CNN has various layers-
1- Convolutional layer – it passes the information to the next layer
2- Pooling layer – it combine the cluster of neuron into single neuron into next layer
3- A fully connected layer – it connect the every neuron in one layer to every neuron in another layer.

CNN learns feature detection through tens or hundreds of hidden layer. Each layer increases the complexity of the learned
features.Various steps in classifying image using CNNFirst it starts with an input image after that many different filters are applied
to create a feature map then ReLU function are applied to increase non-linearity then pooling layer is applied to each feature map
then pooled images are flattened into one long vector. After flattening input the vector into a fully connected artificial neural network
then train it through forward propagation and backward propagation and repeat it until we have a well defined neural network with
trained weights and feature detectors.

III. PROPOSED SYSTEM

Our system will work according to the system architecture shown in the figure below, it will first capture the image through a digital
camera or else it will capture through the database Whether. Each image will be normalized to a predefined size for the next process.
CNN is a multilayer, feed-forward neural networks (FFNN) which can quickly identify, classify, and recognize any features in an
image. It is used mainly with visual data, such as image classification. A CNN can be prepared to do image analysis tasks including
object recognition, segmentation, classification, and image processing. Large-scale image recognition has been become possible
because of large public image databases such as ImageNet. CNN are networks made up of neurons similar to the human brain.
These neurons consists of weights and biases that form layers and fire in a particular order to end up with a final output. The
networks can be trained in order to recognize particular patterns by feeding them large amounts of data. This is very useful in the
field of computer vision since it means that a computer can be trained to recognize different objects.

IJCRT2210371 International Journal of Creative Research Thoughts (IJCRT) www.ijcrt.org d201

www.ijcrt.org © 2022 IJCRT | Volume 10, Issue 10 October 2022 | ISSN: 2320-2882

Tensorflow:
Tensorflow is an open source software for numerical computation. It was originally created to conduct machine learning and deep
neural networks research. Tensorflow provides neural network architectures and scripts to retrain the networks for users who wants
to apply them in different contexts.

Keras:
Keras is an open-source neural-network library written in Python, which is used for the preprocessing, modeling, evaluating, and
optimization. It is capable of running on top of TensorFlow. It is used for high-level API as it handled by backend. It is designed
for making a model with loss and optimizer function, and training process with fit function. For backend, it designed for convolution
and low-level computation under tensors or TensorFlow. Importing the python libraries are used for preprocessing, modelling,
optimization, testing and display emotion which having a maximum percentage.

TABLE 1 Training Dataset

Object Name Number of Training Number of Testing Dataset

Dataset

Apple 162 47
Orange 145 40
Banana 153 43

IJCRT2210371 International Journal of Creative Research Thoughts (IJCRT) www.ijcrt.org d202

www.ijcrt.org © 2022 IJCRT | Volume 10, Issue 10 October 2022 | ISSN: 2320-2882
Result of Classification

IV. FUTURE SCOPE AND CONCLUSION

In this paper we proposed a system that uses convolution neural network for extracting and selecting the features for any given
image and classify the images into appropriate classes. The Convolution neural network can give high accuracy
compared to other classifiers. The performance and accuracy is tested on simple CPU as well as GPU. Hence we conclude that
Convolution Neural Networks are a good choice for Image Classification. Further we will deploy this model in android app and
make this model easy to handle for user.

V. ACKNOWLEDGEMENT

A sincere thanks to Professor Santosh Jadhav for the opportunity to do this project. The amount of knowledge I've gained about
deep learning through this project makes me consider it a \class-within-a-class". Truly, I learned more doing this project than I have
in entire classes.

References

[1] Raja Majid Mehmood 1, Ruoyu Du 2 and Hyo Jong Lee Optimal feature selection and Deep Learning Ensembles Method for
emotion recognition from human brain EEG sensors.
[2] Young Jong Mo, Joongheon Kim, Jong-Kook Kim, Aziz Mohaisen, and Lee,Performance of Deep Learning Computation with
TensorFlow Software Library in GPU-Capable Multi-Core Computing Platforms.
[3] V. A. Bharadi, B. Pandya and B. Nemade, "Multimodal biometric recognition using iris & fingerprint: By texture feature
extraction using hybrid wavelets," 2014 5th International Conference - Confluence The Next Generation Information
Technology Summit (Confluence), Noida, 2014, pp. 697-702. doi: 10.1109/CONFLUENCE.2014.6949309
[4] V. A. Bharadi, P. Mishra and B. Pandya, "Multimodal face recognition using multidimensional clustering on hyperspectral
face images," 2014 5th International Conference - Confluence The Next Generation Information Technology Summit
(Confluence), Noida, 2014, pp. 582-588. doi: 10.1109/CONFLUENCE.2014.6949048

IJCRT2210371 International Journal of Creative Research Thoughts (IJCRT) www.ijcrt.org d203