Birds Species Identification Using Deep Learning

CHAPTER 1

INTRODUCTION

1.1 INTRODUCTION

Birds are the part of nature. They contribute to the ecosystem and environment. Bird
species identification from images is an important and challenging problem with many
applications in the real world. In order to evaluate the quality of our living environment it is
important to obtain reliable information about the population of wild animals. Birds play major
role in Seed propagation, food chain, and the beauty of nature etc. Thus birds are greatly helpful
to humans and nature. Hence the identification of the birds is necessary. Recent machine learning
and deep learning research has put a lot of focus on facial recognition, which involves detecting
faces within images, land marking specific features to account for the angle at which the picture
was taken, and finally determining whose face is in the picture. This is extremely valuable
research and has many applications including social media, computer security, marketing, and
healthcare. While this research is booming, there has been less focus on the classification of
other organisms like birds, for example.

As with facial recognition, there is much complexity in the classification of birds due
issues like the similarities in different subspecies of birds, challenges with the foreground or
background setting of images, and lighting conditions in photos. The present is the Robotic era.
The robots carry pre-defined tasks efficiently with embedded computer vision systems.
Computer vision systems have many applications such as performing object identification, object
recognition etc. The use of automated methods for bird species identification is an effective way
to evaluate the quantity and diversity of the birds. In all these systems, the images are processed
to extract useful features that are necessary for further recognition and classification.

Birds are typically categorized by their shape or color and physical characteristics. Bird
species classification is a challenging problem both to humans and to computational algorithms
because this problem is based on bird’s images. The majority of activities in the real world can
be carried out automatically by the development of robotics. One such application envisaged is
the classification of birds.

Department Of Computer Science, Mandya University, Mandya Page 1

 Birds Species Identification Using Deep Learning

Image processing basically contains the following three steps:

a) Importing the image with ocular scanner or by digital photography.

b) Analyzing and handling the image which includes data condensation and image enhancement
and spotting patterns that are not to human eyes like satellite photographs.

c) Output is the last stage in which result can be changed image or report that is based on image
analysis.

1.2 Motivation
The incitement behind this project is to embellish an alternative approach for
classification of birds because identification and classification is a major challenge for the
computer vision to achieve near human levels of recognition. The bird’s classification is useful
by the forest department and can be utilized in computer vision for the automatic categorizing
the birds from a set, consisting of different kinds of birds. Bird’s play an important roles in the
ecosystem, e.g. providing insect control and seed dispersion and pollination.
1.3 SCOPE OF THE WORK

User will get sufficient knowledge of information about each and every bird species
present in our system. Our proposed application is implemented in order to understand the every
common user without any ambiguity of bird’s description.

1.4 OBJECTIVES

 The primary objective is to develop a robust deep learning model based on the VGG16
architecture for accurately classifying bird species from images.

 Collecting and annotating a comprehensive dataset of bird images representing diverse

species is crucial for training and evaluating the classification model.

 The developed classification model will be rigorously evaluated using standard evaluation
metrics such as accuracy, precision, recall, and F1-score.

Department Of Computer Science, Mandya University, Mandya Page 2

 Birds Species Identification Using Deep Learning

 The final objective is to deploy the trained model into a user-friendly application or
platform; enabling users to upload bird images and receive accurate species identification
results in real-time.

1.5 EXISTING SYSTEM

Gathering and assembling data regarding birds needs large human effort similarly as
becomes awfully costlier methodology. A reliable system which will give massive scale process
of knowledge regarding birds and can function a valuable tool for researchers, governmental
agencies, etc. is needed. So, bird species identification plays a very important role in
distinguishing that a specific image of bird belongs to that species. Bird species identification
suggests that predicting the bird species belongs to that class by victimization a picture. The
identification may be done through image, audio or video. Associate in nursing audio process
technique makes it doable to spot by capturing the audio signal of birds. But, because of the
mixed sounds in surroundings like insects, objects from globe, etc. process of such data becomes
additional difficult.

1.6 PROBLEM STATEMENT

Identification of species requires the assistance of manual bird books. So, it also requires
expertise in the field to identify the species accurately. Few Species of Birds look very familiar
in their appearances thus identifying the exact species by humans may be error prone.
Researchers studying avian ecology, behavior, and migration patterns can benefit from
automated bird species identification systems to efficiently analyze large volumes of bird
observation data.

1.7 PROPOSED SYSTEM

In this project we proposed a methodology for bird’s classification based SVM methods.
We take bird images as training images to train the system first. To do training, initially, pre-
processing techniques will be adopted to convert the RGB image to grayscale image and then
convert grayscale image to binary image using the texture analysis and morphological operations
on digital images of different birds. By performing digital image processing, defined as the
acquisition and processing of visual information by computer, computer vision systems allow

Department Of Computer Science, Mandya University, Mandya Page 3

 Birds Species Identification Using Deep Learning

analyzing image data for specific applications to determine how images can be used to extract
the required information. Then, the preprocessed images are identified by extracting the features
like color moment feature and store in database. Then take one testing image and do the same
preprocessing steps for the testing image, next step is to compare features(values) obtained from
testing bird image with the database values using SVM. Later, the processed images will be
further classified by using SVM and finally get the classified label of bird from classification
methods, which shows which bird belongs to which group in the training sets and finally
compute the accuracy.

FIG: 1.1 General structure of the proposed method.

 Image Loading: Load the bird images into the system using image processing libraries like
open CV or PIL .this step involves reading the image files from the dataset and converting
them into a format that can be used for further analysis.
 Image Preprocessing: The input images of birds are preprocessed to enhance their quality
and prefer them for segmentation .This may include tasks such as resizing, noise reduction,
and contrast enhancement.
 Segmented Image: The preprocessed images are segmented to identify and isolate the
region of interest, such as the birds itself; from the background this can be using various
techniques, such as thresholding , edge detection or clustering algorithms.

Department Of Computer Science, Mandya University, Mandya Page 4

 Birds Species Identification Using Deep Learning

 Features extraction: Once the images are segmented ,features are extracted from the
segmented region .This features may include a shape , texture, color , or other characteristics
that can help distinguish different bird species.
 Neural Network: The neural network architecture, including input layers, hidden layers, and
output layers tailored for image classification. Train the neural network use in the labeled
dataset to learn and recognize patterns specific to each bird species.
 Classification: The extracted features used as input to classification model, such as a
machine learning algorithm or a deep learning model. The model is trained on a label dataset
of segment bird images to learn the patterns and characteristics of different bird species.
 Dataset: Gathering a large dataset of bird images with corresponding species label.
 Bird identification: During the inference phase, the trained model is use to predict the
species of birds in new,segmented images. The model process the extracted features through
layers and outputs a prediction of the bird species present in the images.

Advantages:

 Easy to classify the bird species

 Better accuracy than the existing system
 Time consumption is reduced
 Manpower is reduced greatly

Department Of Computer Science, Mandya University, Mandya Page 5

 Birds Species Identification Using Deep Learning

CHAPTER 2
LITERATURE REVIEW
Each Software development requires the review interaction. The Survey interaction is
expected to get the necessity for the product. The Survey likewise comprises of considering the
current framework and furthermore reading the instruments required for the advancement of the
product. A legitimate comprehension of the apparatuses is a lot of fundamental. Following is a
concentrate of the data of the material gathered during the writing study. A writing review is a
procedure of recognizing the issues in the current framework through research and proposing the
advancement of the framework to take care of the issues of the current framework.

2.1 RESEARCH PAPERS

[1]. Automatic classification of flying birds species using computer vision techniques
Authors : John Atanbori, Wenting Duan
Year : 2018 Findings
The aim of our work is to develop a reliable automated system, capable of classifying the species of
individual birds, during flight, using image data. This is challenging, but appropriate for use in
the field, since there is often a requirement to identify in flight, rather than while stationary.
Which uses a new and rich set of appearance features for classification from image..They also
introduce motion features including curvature and wing beat frequency. Combined with Normal
Bayes classifier and a Support Vector Machine classifier, They present experimental evaluations
of our appearance and motion features across a data set comprising 7 species. Using appearance
feature set alone they achieved a classification rate of 92% and 89% (using Normal Bayes and
SVM classifiers respectively) which significantly outperforms a recent comparable state-of-the-
art system.
Cons:
The system's performance may be affected by variations in environmental conditions, lighting, bird
movements, and background clutter, potentially leading to reduced classification accuracy in
real-world scenarios.

Department Of Computer Science, Mandya University, Mandya Page 6

 Birds Species Identification Using Deep Learning

[2]. Content-Based Bird Retrieval using Shape context, Color moments and Bag
of Features
Authors : Bahri abdelkhalak

Year : 2019

Findings

In this paper we propose a new descriptor for birds search. First, our work was carried on
the choice of a descriptor. This choice is usually driven by the application requirements such as
robustness to noise, stability with respect to bias, the invariance to geometrical transformations
or tolerance to occlusions. In this context, we introduce a descriptor which combines the shape
and color descriptors to have an effectiveness description of birds. The proposed descriptor is an
adaptation of a descriptor based on the contours defined in article Belongie et al. combined with
color moments. Specifically, points of interest are extracted from each image and information’s
concatenated with color moments. Thus, the approach bag of visual words is applied to the latter.
The experimental results show the effectiveness of our descriptor for the bird search by content.
This experiment achieves the accuracy of 82%.

Cons:
The combination of shape and color descriptors may increase the complexity of the descriptor,
potentially requiring higher computational resources and limiting real-time application in
resource-constrained environments.

[3]. An Exploration of computer vision techniques for bird species classification

Authors: Anne L. Alter, Karen M. Wang
Year : 2018
Findings
Bird classification, a fine-grained categorization task, is a complex task but crucial in improving
and identifying the best computer vision algorithms to use in the broader image recognition field.
Di_culties like lighting conditions, complex foliage settings, and imilarities in subspecies of
birds are just some of the challenges faced by researchers. We implemented softmax regression
on manually observed binary attributes, a multi-class SVM on HOG and RGB features from

Department Of Computer Science, Mandya University, Mandya Page 7

 Birds Species Identification Using Deep Learning

photos, and finally a CNN using transfer learning to classify birds. The pre-trained CNN with
fixed feature extraction proved to be the best method for classification with computer vision.
Cons;
The complexity of CNNs and transfer learning may require significant computational resources
for training and inference Variations in lighting conditions, backgrounds, and bird poses can
affect the performance of classification algorithms, potentially leading to reduced accuracy in
real-world scenarios.

[4].Bird species classification based on color features

Authors : Andreia Marini, Jacques Facon
Year : 2019
Findings
Proposed a work of bird species classification based on color features extracted from
unconstrained images. This means that the birds may appear in different scenarios as well may
present different poses, sizes and angles of view. Besides, the images present strong variations in
illuminations and parts of the birds may be occluded by other elements of the scenario. The
proposed approach first applies a color segmentation algorithm in an attempt to eliminate
background elements and to delimit candidate regions where the bird may be present within the
image. Next, the image is split into component planes and from each plane, normalized color
histograms are computed from these candidate regions. After aggregation processing is
employed to reduce the number of the intervals of the histograms to a fixed number of bins. The
histogram bins are used as feature vectors to by a learning algorithm to try to distinguish between
the different numbers of bird species.
Cons:
The accuracy of the color segmentation algorithm may vary depending on the complexity of the
background, lighting conditions, and occlusions Using normalized color histograms as feature
vectors may not capture all relevant information for distinguishing between bird species,
particularly in scenarios with significant variations in bird appearance and pose.

[5]. Classification of birds based on their sound patterns using GMM and SVM classifiers
Authors : Supriya, Prakrithi R

Department Of Computer Science, Mandya University, Mandya Page 8

 Birds Species Identification Using Deep Learning

Year : 2021

Findings

This paper classifies the bird automatically considering their chirping patterns. Speech
processing and Machine learning techniques are carried out to categorize the bird species. An
attempt has been made to include the bird’s local to South Canara region of India in order to
understand the ecological interdependence. Gaussian Mixture model (GMM) and Support Vector
Machine (SVM) are employed here to classify the birds into the respective classes. An overall
efficiency of 95% is gained using 50 recordings of 4 bird classes using GMM and an efficiency
of 86% through 50 recordings of 5 bird classes using SVM. This project is helpful for
Ornithologists for identification of birds on hearing their chirps.
Cons:
Chirping patterns may vary significantly within and between bird species due to factors such as
age, sex, environmental conditions, and individual variations
The classification model may only be applicable to the bird species local to the South Canara
region of India, limiting its generalizability to other geographic regions and bird populations

[6]. Classification of Bird Species using Audio processing and Deep Neural Network
Authors : Sachin Aggarwal, Smriti Sehgal

Year : 2022

Findings

This paper explores bird species classification through sound analysis, crucial for ornithologists
and birders in tracking rare species. Various techniques such as VGG16, ResnetV2, Multilayer
Perceptron, Naïve Bayes, and Decision trees, alongside audio feature extractors like
spectrograms and Mel-frequency cepstral coefficients, are discussed. Preprocessing techniques
including file conversion and class balancing are employed. The study proposes a Deep
Learning-based classification model to analyze bird sounds, aiming to streamline species
identification and conservation efforts.
Cons:

Department Of Computer Science, Mandya University, Mandya Page 9

 Birds Species Identification Using Deep Learning

The accuracy of classification may be influenced by the quality and variability of the audio data,
including background noise, recording conditions
The classification model may only be applicable to the species and environments included in the
training dataset, limiting its generalizability to other geographic regions.

[7]. Design of Bird Sound Recognition Model Based on Lightweight

Authors : Fan Yang, Ying Jiang

Year : 2022

Findings
This paper presents a lightweight bird sound recognition model using MobileNetV3 as the
backbone. A large dataset comprising 264 bird species enhances model generalization.
Depthwise separable convolution and multi-scale feature fusion improve recognition ability. The
introduction of the Pyramid Split Attention (PSA) module enhances adaptability to spatial and
channel information extraction. Additionally, the Bneck module is refined with channel attention
and ordinary convolution. Experimental results show improved accuracy compared to
MobileNetV1, MobileNetV2, and MobileNetV3, with significantly fewer parameters and
FLOPs.
Cons:
Despite the use of a large dataset, the model's generalization ability may still be limited to the
bird species included in the training dataset
The effectiveness of the model relies on the extraction of intricate features from bird sound data,
which may be challenging to capture accurately

[8]. Machine Learning Approach to Classify Birds on the Basis of Their Sound
Authors : Yogesh Jadhav, Vishal Patil

Year : 2020

Findings
The Bird Classifier system employs machine learning to identify bird species based solely on
recorded bird sounds. It preprocesses recorded datasets to remove noise and extract useful

Department Of Computer Science, Mandya University, Mandya Page 10

 Birds Species Identification Using Deep Learning

features like MFCC and Mel-Spectra. Various classification algorithms including KNN, Random
Forest, Multi-layer Perceptron, and Bayes are tested, with the most effective algorithm chosen
for implementation. The system facilitates the translation of annotations into datasets for
training, enabling efficient species identification and presence/absence detection.
Cons:
Despite using advanced machine learning algorithms, the accuracy of bird species classification
may vary depending on the quality of the recorded sound data The effectiveness of the system
heavily relies on the quality and representativeness of the training data, and biases or
inaccuracies in the dataset may impact the model's performance and generalization ability.

[9].Classify bird species audio by augument convolutional neural network

Authors : Hasan Abdullah Jasim, Saadaldeen R. Ahmed

Year : 2022

Findings
This thesis proposes a fully automated bird species identification system using convolutional
neural networks (CNNs) based on spectrogram images. By analyzing challenges in bird species
detection, segmentation, and classification, CNNs with Fully Convolutional Learning are
employed for efficient results. Leveraging a subset of 8000 sound samples, the system achieves a
loss of less than 0.0063 and an accuracy of 0.9895. Through a 9-step implementation, bird
species spectrograms can be detected from audio samples with high precision and efficiency.
Cons:
The accuracy of species identification may be influenced by the quality and variability of the
spectrogram images, including variations in recording conditions
The effectiveness of the system may be limited to the bird species included in the dataset used
for training and validation, and its performance may degrade when applied to unseen species or
environments not represented in the training data.

[10]. Bird Sound Identification based on Artificial Neural Network

Authors : Muhammad Mujjajdid Mohd Sukri, Umi Fadlilah

Year : 2020

Department Of Computer Science, Mandya University, Mandya Page 11

 Birds Species Identification Using Deep Learning

Findings
This paper proposes a bird sound identification system using Artificial Neural Network (ANN)
to aid researchers in species recognition. MATLAB software is utilized to process power spectral
density data for each bird type and train the ANN for species classification. A graphical user
interface (GUI) is developed to allow users to input bird audio for species recognition. The
system successfully identifies bird species based on their unique sounds, providing a valuable
tool for wildlife researchers.
Cons:
The accuracy of species identification may vary depending on the quality of the audio input,
background noise, and the complexity of bird vocalizations
The system may only be able to identify one bird species at a time, potentially limiting its utility
in scenarios where multiple bird species are present.

2.2 SUMMARY OF LITERATURE

A literature survey on bird species identification involves reviewing existing research

and studies on the various methods and technologies used to identify different bird species.
Reasearches typically examine the strengths and limitations of techniques such as field
observations and image recognition in accurately identifying bird species. The survey aims to
provide a comprehensive overview of the current state of research in bird species identification,
highlighting key findings, challenges areas for further investigation. By synthesizing information
from multiple sources, researches can gain deeper understanding of methods and technologies
employed in bird species identification and contribute to the advancement of this field.if you
require more specific information or references.

Department Of Computer Science, Mandya University, Mandya Page 12

 Birds Species Identification Using Deep Learning

CHAPTER 3
SYSTEM REQUIREMENTS
The Software Requirement Specification (SRS) is fundamental data, which shapes the
foundation of the product advancement measure. SRS records the necessities of a system just as
has a portrayal of its critical segments. The concentration in this stage is one of the clients of the
framework and not the framework arrangements. The consequence of the necessity particular
archive expresses the expectation of the product, properties, and imperatives of the ideal
framework. SRS comprises the comprehension among clients and creators concerning the
substance of the item that will be made. SRS ought to be exact and thoroughly imply the
structure requirements as it makes a monster obligation to the overall advancement plan.

Perhaps the most fundamental snippets of data are SRS (Software Requirement
Specification). It gives definite data about the foundation of the product advancement measure. It
records the significant necessities of the structure likewise holds the portrayal of the significant
segments. These things will be in the IEEE guidelines. The proposals would shape the
clarification behind giving an unmistakable picture of the thing to be made filling in as an action
for the execution of comprehension among the customer and the designer. One of the significant
advances associated with the improvement cycle is framework prerequisites. This SRS (Software
Requirement Specification) is trailed in the asset investigation stage. Its fundamental errand is to
choose what a product item does. In this stage, the primary center is the client and not the
framework arrangement. SRS (Software Requirement Specification) gives the outcomes like the
goal of the product, properties, and limitations of the ideal framework. The primary benefit of
SRS (Software Requirement Specification) is that it gives an unmistakable comprehension
among the customers and the item engineers as for the item that is created. SRS (Software
Requirement Specification) which is archived should precise and the essentials of the system
ought to be implied as it makes a huge obligation to the overall advancement plan measure.

3.1 Functional Requirements

Functional requirements specify what a system should do. They describe the expected
behavior of the system, detailing the functionality, tasks, and services it must provide. These

Department Of Computer Science, Mandya University, Mandya Page 13

 Birds Species Identification Using Deep Learning

requirements focus on user interactions, the system's inputs and outputs, and how the system
processes information.

 Load Data : The images of the birds are captured through the camera, this image is in RGB
(Red, Green and Blue) form, color transformation structure for the bird image is created,
and then an independent color space transformation for the color transformation structure is
applied.
 Pre-processing:The aim of preprocessing is an improvement of the image data that
suppresses unwanted distortions or enhance some image feature important for further
processing.
 Feature extraction:The aim of feature extraction is find out and extract features that can be
used to determine the meaning of given sample.
 Classification:In this phase to detect and classify the bird species, we are using the
classifier that is support vector machine.

3.2 Non-functional Requirements

Non-functional requirements (NFRs) specify how a system should perform rather than
what it should do. They define the quality attributes, constraints, and performance metrics of the
system, addressing the operational aspects that affect user experience and system effectiveness.

 Performance Requirements: The product should uphold the end-clients necessities. The
product is equipped for handling when the enormous quantities of records are given as
information and furthermore, it should be intelligent and the deferrals included ought to be
less. So in each activity reaction of the framework, there are no prompt deferrals.
 Safety and Security Requirements: The system ought to be planned in a safe manner
by applying wellbeing measures. Data transmission ought to be safely communicated
to hubs with no progressions in data. A unique special case dealing with components
ought to be set up to keep away from system mistakes.
Software Quality Attribute Availability:
The application will not hang and opens quickly and with 99.9% uptime.
Reliability:

Department Of Computer Science, Mandya University, Mandya Page 14

 Birds Species Identification Using Deep Learning

The system should not crash and should identify invalid input and produce suitable error
message.
Usability:
The interface should be intuitive and easily navigable and user friendly.
Integrity:
The software does not store any cache data or doesn’t use system resources in
background.

3.3 FEASIBILITY STUDY

The feasibility study permits discovering answers for the issues of the undertaking. The
arrangement is given what how resembles a fresh out of the new device resembles.

TECHNICAL FEASIBILITY

The software's utilized for implementing for the clients to move the insights in cross breed
organizations. Without the race condition issues and invalid reservations, the data needs to move
from source to objective. To keep away from the channel fluctuation and transfer speed confined
to change measurements without issues to the objective.

ECONOMIC FESIBILITY

The general expense of the venture could be truly reasonable as we have utilized open-source
programming like python. The investigation additionally assisted the organization with
administering the sum that would be brought about before an undertaking is in the end support.
This permitted organization to admirably achieve their cash to ensure the most productive
venture is attempted.

Economically the Scalability application is feasible as the developmental and maintenance

charges are less than 15000 per year.

OPERATIONAL FEASILBILITY

The task will be utilized by the association under various conditions. Anybody can work with
this application as it upholds an easy-to-use approach. It gives graphical UIs to the client so the
client can without much of a stretch connect with the framework. Clients no compelling reason

Department Of Computer Science, Mandya University, Mandya Page 15

 Birds Species Identification Using Deep Learning

to know about python and IDLE ideas to utilize the application. The application is planned so
that it tends to be effectively carried out in any form of OS.

SCHEDULE FEASIBILITY

The time schedule required for the development of this project is 5 months and it is very
important since over-runs result in escalated projects costs and also hinders in the development
of the other systems.

20 days – Requirement gathering

25 days – System designing

55 days – Implementation

35 days – Testing and deployment

3.4 TOOLS AND TECHNOLOGY

1. Python

Python is an interpreted, high-level, general-purpose programming language. Created by

Guido van Rossum and first released in 1991, Python's design philosophy emphasizes code
readability with its notable use of significant whitespace. Its language constructs and object-
oriented approach aim to help programmers write clear, logical code for small and large-scale
projects.Python was conceived in the late 198os as a successor to the ABC language. Python 2.o,
released in 2ooo, introduced features like list comprehensions and a garbage collection system
capable of collecting reference cycles. Python 3.o, released in 2oo8, was a major revision of the
language that is not completely backward-compatible, and much Python 2 code does not run
unmodified on Python 3.Python interpreters are available for many operating systems. A global
community of programmers develops and maintains CPython, an open source reference
implementation. A non-profit organization, the Python Software Foundation, manages and
directs resources for Python and CPython development.

2. OpenCV

Department Of Computer Science, Mandya University, Mandya Page 16

 Birds Species Identification Using Deep Learning

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

aimed at real-time computer vision. Originally developed by Intel, it was later supported by
Willow Garage then Itseez (which was later acquired by Intel). The library is cross-platform and
free for use under the open-source BSD license. OpenCV supports some models from deep
learning frameworks like TensorFlow, Torch, PyTorch (after converting to an ONNX model)
and Caffe according to a defined list of supported layers. It promotes OpenVisionCapsules,
which is a portable format, compatible with all other formats.

Officially launched in 1999 the OpenCV project was initially an Intel Research initiative to advance
CPU-intensive applications, part of a series of projects including real-time ray tracing and 3D
display walls. The main contributors to the project included a number of optimization experts in
Intel Russia, as well as Intel's Performance Library Team.

3. Tensorflow

TensorFlow is a free and open-source software library for dataflow and differentiable
programming across a range of tasks. It is a symbolic math library, and is also used for machine
learning applications such as neural networks. It is used for both research and production at
Google.TensorFlow was developed by the Google Brain team for internal Google use. It was
released under the Apache License 2.o on November 9, 2o15.TensorFlow is Google Brain's
second-generation system. Version 1.o.o was released on February 11, 2o17. While the reference
implementation runs on single devices, TensorFlow can run on multiple CPUs and GPUs (with
optional CUDA and SYCL extensions for general-purpose computing on graphics processing
units). TensorFlow is available on 64-bit Linux, macOS, Windows, and mobile computing
platforms including Android and iOS. Its flexible architecture allows for the easy deployment of
computation across a variety of platforms (CPUs, GPUs, TPUs), and from desktops to clusters of
servers to mobile and edge devices.

4. Pandas

Pandas is an open source Python package that is most widely used for data science/data
analysis and machine learning tasks. It is built on top of another package named Numpy, which
provides support for multi-dimensional arrays. As one of the most popular data wrangling

Department Of Computer Science, Mandya University, Mandya Page 17

 Birds Species Identification Using Deep Learning

packages, Pandas works well with many other data science modules inside the Python
ecosystem, and is typically included in every Python distribution, from those that come with
your operating system to commercial vendor distributions like Active State’s Active Python.

5 Numpy
NumPy is a library for the Python programming language, adding support for large,
multi-dimensional arrays and matrices, along with a large collection of high-
level mathematical functions to operate on these arrays. The ancestor of NumPy, Numeric, was
originally created by Jim Hugunin with contributions from several other developers. In
2005, Travis Oliphant created NumPy by incorporating features of the competing Numarray into
Numeric, with extensive modifications. NumPy is open-source software and has many
contributors. The Python programming language was not originally designed for numerical
computing, but attracted the attention of the scientific and engineering community early on. In
1995 the special interest group (SIG) matrix-sig was founded with the aim of defining
an array computing package; among its members was Python designer and maintainer Guido van
Rossum, who extended Python's syntax (in particular the indexing syntax) to make array
computing easier. An implementation of a matrix package was completed by Jim Fulton, then
generalized by Jim Hugunin and called Numeric (also variously known as the "Numerical
Python extensions" or "NumPy"). Hugunin, a graduate student at the Massachusetts Institute of
Technology (MIT), joined the Corporation for National Research Initiatives (CNRI) in 1997 to
work on JPython, leaving Paul Dubois of Lawrence Livermore National Laboratory (LLNL) to
take over as maintainer.

3.5 HARDWARE REQUIREMENTS

 Processor : Intel i3 2.53GHz

 Hard Disk : 40GB
 RAM : 4 GB or above

3.6 SOFTWARE REQUIREMENTS

 Operating system : Windows 8 and above

 Coding Language : Python
 IDE : IDLE

Department Of Computer Science, Mandya University, Mandya Page 18

 Birds Species Identification Using Deep Learning

CHAPTER 4

SYSTEM DESIGN

This system contains the system design in the detailed design starts after the system
design phase is completed and the system design has been certified through the review. The goal
of this phase is to develop the internal logic of each of the modules identified during system
design.

In the system design, the focus is on identifying the modules, whereas during detailed
design the focus is on designing the logic for the modules. In other words in system design
attention is on what components are needed, while in detailed design how the components can be
implemented in the software is the issue.

The design activity is often divided into two separate phase system design and detailed
design. System design is also called top-level design. At the first level focus is on deciding
which modules are needed for the system, the specifications of these modules and how the
modules should be interconnected. This is called system design or top level design. In the second
level the internal design of the modules or how the specifications of the module can be satisfied
is decided. This design level is often called detailed design or logic design.

4.1 USE CASE DIAGRAM

Use case diagram is a graph of actors, a hard and fast of use instances enclosed by means
of a device boundary, conversation associations among the actor and the use case. The use case
diagram describes how a gadget interacts with out of doors actors; each use case represents a bit
of functionality that a machine provides to its users. A use case is called an ellipse containing the
call of the use case and an actor is shown as a stick figure with the call of the actor beneath the
parent.

The use instances are used at some point of the evaluation phase of a task to pick out and
partition system capability. They separate the device into actors and use case. Actors represent
roles which might be played by using person of the system. Those users may be people, different
computer systems, portions of hardware, or maybe other software structures.

Department Of Computer Science, Mandya University, Mandya Page 19

 Birds Species Identification Using Deep Learning

Bird Species
Identification

Image Input

Border
Selection

Border
Histogram

Color
Removal
User

Segmented
Image

Classificatio
n

Outpu
t

Fig 4.1 Use case diagram for User

In the above user use case diagram, First user will provide an image as input to the
system. Once the system receives the input then it will do the border selection form the given
image and border histogram will be done. After this color removal from the image will be done
and image segmentation will take place and classification of bird species will be given as output
to the user.

4.2 DATA FLOW DIAGRAM

DFD graphically representing the functions, or processes, which capture, manipulate,

store, and distribute data between a system and its environment and between components of a
system. The visual representation makes it a good communication tool between User and

Department Of Computer Science, Mandya University, Mandya Page 20

 Birds Species Identification Using Deep Learning

System designer. Structure of DFD allows starting from a broad overview and expand it to a
hierarchy of detailed diagrams. DFD has often been used due to the following reasons:
 Logical information flow of the system
 Determination of physical system construction requirements
 Simplicity of notation
 Establishment of manual and automated systems requirements

Data Flow Diagram

Image
Inputt

Bird Species
Identification

Prediction
of Bird

Level 0 Data Flow Diagram

In the above level 0 dfd, the system will be given an image as input and the system will
predict the species of the bird based on the input.

In the below level 1 user dfd, User will provide an image input then once the system
receives the input then it will do the border selection form the given image and border histogram
will be done. After this color removal from the image will be done and image segmentation will
take place and classification of bird species will be given as output to the user.

Department Of Computer Science, Mandya University, Mandya Page 21

 Birds Species Identification Using Deep Learning

Border Border
Use Image
Selectio Histogra
n m

Segmente
Outpu Classificatio Color
d Image

Level 1 Data Flow Diagram – User

4.3 ACTIVITY DIAGRAM

In this developed project the Activity diagrams illustrate the overall flow of control. This
diagram symbolizes the goings-on taking place in the project. There are different
accomplishments for member.

Basic Notations

Initial Activity

This shows the starting point or first activity of the flow. It is denoted by a solid circle.

Final Activity

The end of the Activity diagram is shown by a bull's eye symbol, also called as a final
activity.

Activity

Department Of Computer Science, Mandya University, Mandya Page 22

 Birds Species Identification Using Deep Learning

Represented by a rectangle with rounded (almost oval) edges

Image
Input

Border
Selection

Border
Histogra

Color
Removal

Segmente
d

Classificatio
n

Outp
ut

Fig 4.3 Activity diagram for User

Department Of Computer Science, Mandya University, Mandya Page 23

 Birds Species Identification Using Deep Learning

In the above user activity diagram, User will load an image as input to the system and then
system will perform border selection on the image, border histogram, color removal and
segmented image and classification will be done and user can view the output.

4.4 SEQUENCE DIAGRAMS:

A sequence diagram shows object interactions arranged in time sequence. It depicts the
objects and classes involved in the scenario and the sequence of messages exchanged between
the objects needed to carry out the functionality of the scenario. Sequence diagrams are
sometimes called event diagrams, event scenarios.UML sequence diagrams are used to represent
or model the flow of messages, events and actions between the objects or components of a
system. Time is represented in the vertical direction showing the sequence of interactions of the
header elements, which are displayed horizontally at the top of the diagram Sequence Diagrams
are used primarily to design, document and validate the architecture, interfaces and logic of the
system by describing the sequence of actions that need to be performed to complete a task or
scenario.

User System:

Image
Input

Border
Selection

Border
Histogra
m

Segmente
d Image

Classificatio

Outpu
t

Fig 4.4 Sequence diagram for User

Department Of Computer Science, Mandya University, Mandya Page 24

 Birds Species Identification Using Deep Learning

CHAPTER 5

METHODOLOGY

Implementation is the way toward changing over another or an updated framework plan
into an operational one. The goal is to put the new or overhauled framework that has been tried
into activity while holding expenses, dangers, and individuals bothering to the base. A basic part
of the execution cycle is to guarantee that there will be no disturbing the working of the
association. The best strategy for acquiring control while embedding any new framework is to
utilize all-around arranged tests for testing every single new program. Before creation documents
are utilized to test live information, text records should be made on the old framework, replicated
over to the new framework, and utilized for the underlying trial of each program.

Another factor to be considered in the execution stage is the obtaining of the equipment
and programming. When the product is created for the framework and testing is completed, it is
then the way toward making the recently planned framework completely operational and reliable
in execution.

Implementation is the most significant stage in accomplishing a fruitful framework and

giving the client certainty that the new framework is useful and viable. Execution of an altered
application to supplant a current one. This sort of discussion is moderately simple to deal with,
give there are no significant changes in the framework.

5.1 DATASET

A Dataset is a collection of data, typically organized in a structured format, which can be

used for analysis, research, or training machine learning models. Datasets can vary widely in
size, type, and complexity, and can include various forms of data such as numerical values, text,
images, audio, and video.

Department Of Computer Science, Mandya University, Mandya Page 25

 Birds Species Identification Using Deep Learning

Original Image Grayscaling Image Threshold Image Birds Name

& Accuracy

1 Blue Jay

95.6%

Sparrow
2
94.9%

3 Peacock

95%

4 Owl

94.8%

Department Of Computer Science, Mandya University, Mandya Page 26

 Birds Species Identification Using Deep Learning

Bat
5
95.14%

6
Weaver

95.3%

7 Spoon Bill

95%

Department Of Computer Science, Mandya University, Mandya Page 27

 Birds Species Identification Using Deep Learning

Crow

8 94.57%

9
Turkey

95%

1
Sparrow
0

94.9%

FIG 5.1 :DATASET

1 . Data Collection:

First, a large dataset of bird images and corresponding species labels is collected. This
dataset is crucial for training the identification model. Collect a diverse set of bird images with
different species, angles, lighting conditions, and backgrounds. Ensure the images are labeled
with the correct bird species for supervised learning.

2. Preprocessing:

The images are preprocessed to standardize their size, color, and orientation. This step
ensure consistency in the data fed into the model. Resize, crop, and normalize the images ensure

Department Of Computer Science, Mandya University, Mandya Page 28

 Birds Species Identification Using Deep Learning

To uniformity in size and quality. This step helps in enhancing the model's performance.

3. Feature Extraction:

Features like shapes, colors, textures, and patterns are extracted from the images. These
features help the model distinguish between different bird species. Feature extraction for bird
species identification using deep learning involves automatically identifying and extracting
relevant characteristics from bird images to classify them into different species. This process
leverages the capabilities of convolutional neural networks (CNNs) and other deep learning
models to discern patterns, textures, shapes, and colors that are unique to specific birds species.

4. Model Training:

Machine learning algorithms, such as convolutional neural networks (CNNs), are trained
on the preprocessed data. During training, the model learns to associate specific features with
different bird species. Training a deep learning model for bird species identification involves
several key steps: data preparation, model selection and configuration, training, and evaluation.
Below is a step-by-step guide to accomplish this using a convolutional neural network (CNN).

5. Validation and Testing:

The trained model is validated using a separate dataset to ensure its accuracy. Testing
involves feeding new, unseen images to the model to evaluate its performance. Validation and
testing are crucial steps in the deep learning model training pipeline for bird species
identification. They help ensure that the model generalizes well to unseen data and performs
robustly across different datasets.

6. Deployment:

Once the model is trained and validated, it can be deployed in various applications like
bird identification apps or conservation projects. Deploying a deep learning model for bird
species identification involves making the trained model available for use in a production
environment. This includes setting up an interface (such as a web or mobile application) where
users can upload images of birds, and the model will identify the species.

Department Of Computer Science, Mandya University, Mandya Page 29

 Birds Species Identification Using Deep Learning

7. Continuous Improvement:

The model can be continuously improved by retraining it with new data to enhance its
accuracy and ability to identify bird species more effectively. Continuous improvement of a deep
learning model for bird species identification involves regular updates and enhancements to the
model to maintain and improve its performance over time. This can be achieved through a
combination of monitoring, retraining with new data, refining the model architecture, and
leveraging advanced techniques.

8. Data Acquisition:

The action of retrieving n image from source for processing is known as data
acquisition. It is the first step in the sequence because without an image, no process is possible.
In this process we are considering RGB color images of bird’s as input image. We captured these
images from camera to get more clarity of color pixels.
9. Image resizes process:
Image resizing is the process of changing the size of the image; here pixel values are
also changed. After resizing the image, will not blur and quality of the image will not change.
The original image will be resized into the standard image size 256*256. This standard size
image can reduce the complexity of the image, it reducing dimensionality of an original image.
So, this resized image will be easier for the next process.

10. Conversion of resized image into gray scale image:

In this stage, resized RGB color images will be converted to gray scale image,
because the intensity of gray scale images is less than the color image. Gray scale
image is also known as intensity, gray scale or gray level image. Gray scale is to
reduce the complexity of an image.

11. Conversion of gray scale image into binary image:

In this stage we are converting gray scale image in to Black and White image,
images are composed of pixels and in binary image every pixel value is either 0 or 1 i.e., either
black or white, but in gray scale image can have any value between 0 to 255 for 8-bit color
(every pixel is represented by 8 bits)it can have transition between pure black and pure white.

Department Of Computer Science, Mandya University, Mandya Page 30

 Birds Species Identification Using Deep Learning

12. Segmenting the binary image:

In this stage the image is segmented. According to the region of interest to extract the
feature the image is cropped, and then based on the color histogram the images are get
segmented.
13. Classification

In this phase bird’s species will be identified with the help of SVM. Classification in bird
species identification using deep learning involves training a neural network model to categorize
input images of birds into predefined species categories. This process leverages the ability of
deep learning, particularly convolutional neural networks (CNNs), to automatically learn and
extract relevant features from images.

 SVM

Support Vector Machine (SVM) is a supervised machine learning algorithm which can be used
for both classification or regression challenges. However, it is mostly used in classification
problems. In the SVM algorithm, we plot each data item as a point in n-dimensional space
(where n is number of features you have) with the value of each feature being the value of a
particular coordinate. Then, we perform classification by finding the hyper-plane that
differentiates the two classes very well .

Department Of Computer Science, Mandya University, Mandya Page 31

 Birds Species Identification Using Deep Learning

Support Vectors are simply the co-ordinates of individual observation. The SVM classifier is a
frontier which best segregates the two classes (hyper-plane/ line).

How does it work?

 Identify the right hyper-plane (Scenario-1): Here, we have three hyper-planes (A, B
and C). Now, identify the right hyper-plane to classify star and circle.

You need to remember a thumb rule to identify the right hyper-plane: “Select the hyper-
plane which segregates the two classes better”. In this scenario, hyper-plane “B”
has excellently performed this job.

 Identify the right hyper-plane (Scenario-2): Here, we have three hyper-planes (A, B
and C) and all are segregating the classes well. Now, How can we identify the right
hyper-plane?
 By finding the optimal hyperplane, machine learning models can accurately distinguish
between various bird species. Techniques like Support Vector Machines (SVM) are
commonly used to identify the right hyperplane by maximizing the margin between
different species, leading to improved classification accuracy.

Department Of Computer Science, Mandya University, Mandya Page 32

 Birds Species Identification Using Deep Learning

Here, maximizing the distances between nearest data point (either class) and hyper-plane will
help us to decide the right hyper-plane. This distance is called as Margin.

Above, you can see that the margin for hyper-plane C is high as compared to both A and
B. Hence, we name the right hyper-plane as C. Another lightning reason for selecting the
hyper-plane with higher margin is robustness. If we select a hyper-plane having low
margin then there is high chance of miss-classification.

 Identify the right hyper-plane (Scenario-3):Hint: Use the rules as discussed in previous
section to identify the right hyper-plane

Department Of Computer Science, Mandya University, Mandya Page 33

 Birds Species Identification Using Deep Learning

Some of you may have selected the hyper-plane B as it has higher margin compared to A. But, here
is the catch, SVM selects the hyper-plane which classifies the classes accurately prior to
maximizing margin. Here, hyper-plane B has a classification error and A has classified all
correctly. Therefore, the right hyper-plane is A.

 Can we classify two classes (Scenario-4)?: Below, I am unable to segregate the two
classes using a straight line, as one of the stars lies in the territory of other(circle) class as
an outlier.

Department Of Computer Science, Mandya University, Mandya Page 34

 Birds Species Identification Using Deep Learning

As I have already mentioned, one star at other end is like an outlier for star class. The
SVM algorithm has a feature to ignore outliers and find the hyper-plane that has the
maximum margin. Hence, we can say, SVM classification is robust to outliers.

 Find the hyper-plane to segregate to classes (Scenario-5): In the scenario below, we

can’t have linear hyper-plane between the two classes, so how does SVM classify these
two classes? Till now, we have only looked at the linear hyper-plane.

Department Of Computer Science, Mandya University, Mandya Page 35

 Birds Species Identification Using Deep Learning

SVM can solve this problem. Easily! It solves this problem by introducing additional
feature. Here, we will add a new feature z=x^2+y^2. Now, let’s plot the data points on
axis x and z:

In above plot, points to consider are:

o All values for z would be positive always because z is the squared sum of both x
and y
o In the original plot, red circles appear close to the origin of x and y axes, leading
to lower value of z and star relatively away from the origin result to higher value
of z.

 LIBRARIES AND PACKAGES

Imports

The code starts by importing necessary libraries and modules that will be used throughout
the program.

 OpenCV: A popular computer vision library for image processing and feature extraction.

Department Of Computer Science, Mandya University, Mandya Page 36

 Birds Species Identification Using Deep Learning

 TensorFlow: An open-source machine learning framework for building and training

deep learning models.
 PyTorch: Another deep learning library that provides flexibility and speed for model
training.
 Librosa: A Python package for audio analysis and feature extraction from sound
recordings.
 Scikit-learn: A machine learning library that offers tools for data preprocessing, model
evaluation, and classification tasks.

CLASSES

1. DataLoader Class:

Create a class named DataLoader to handle loading and preprocessing the bird image dataset.

 Include methods such as load_data() to load the images and preprocess_data() to resize
and normalize the images.

2. Model Class:

 Define a class called Model to build the neural network for bird species identification.
 Implement methods like build_model() to define the model architecture using
TensorFlow and Keras.
 Include functions for model compilation, training, evaluation, and prediction.

3. Bird Species Identification Class:

 Develop a class named BirdSpeciesIdentification as the main interface for the model
and dataset.
 Instantiate objects from the DataLoader and Model classes within this class.
 Implement methods like train_model() to train the model, evaluate_model() to assess
its performance, and predict_species() for making predictions.

Department Of Computer Science, Mandya University, Mandya Page 37

 Birds Species Identification Using Deep Learning

CHAPTER 6

EXPERIMENTAL ANALYSIS AND RESULT

Conducting an experimental analysis for bird species identification typically involves

several steps, including data collection, preprocessing, model training, evaluation, and result
interpretation.

6.1 EXPERIMENTAL ANALYSIS

FIG 6.1 :SAMPLE EXPERIMENT ANALYSIS

Department Of Computer Science, Mandya University, Mandya Page 38

 Birds Species Identification Using Deep Learning

 In bird species identification using images, one of the key technologies employed
is computer vision. Computer vision algorithms are used to analyze visual
features in bird images to distinguish between different species.
 One popular approach is the use of convolutional neural networks (CNNs), a type
of deep learning algorithm that is effective in image recognition tasks.
Researchers can train CNN models on large datasets of bird images to learn the
unique visual patterns associated with each species.
 Additionally, transfer learning is often used in bird species identification.
Researchers can leverage pre-trained CNN models (like ResNet, VGG, or
Inception) on general image datasets and fine-tune them on bird-specific datasets
to improve the accuracy of species identification.
 Furthermore, image processing techniques such as edge detection, feature
extraction, and segmentation can be utilized to enhance the quality of images
before feeding them into the identification models.
 By combining these technologies and methods, researchers can develop robust
bird species identification systems that can accurately classify bird species based
on images.

6.2 RESULT

In bird species identification using deep learning, the results often show significant
advancements in accuracy and efficiency compared to traditional methods. Deep learning
models, especially convolutional neural networks (CNNs), have demonstrated impressive
capabilities in recognizing complex patterns in bird images, leading to more accurate species
identification.

The results of bird species identification using deep learning typically include high
accuracy rates, with some studies achieving accuracy levels above 90% in classifying bird
species from images. These deep learning models can effectively learn intricate features and
characteristics of different bird species, enabling them to make accurate predictions. Researchers
also evaluate the performance of deep learning models using metrics such as precision, recall,

Department Of Computer Science, Mandya University, Mandya Page 39

 Birds Species Identification Using Deep Learning

and F1 score to assess the model's ability to correctly identify bird species. These metrics
provide insights into the model's strengths and areas for improvement.

Overall, deep learning has shown great promise in bird species identification, offering
advanced capabilities in image analysis and classification. The results of studies using deep
learning for bird species identification highlight the potential for accurate and efficient species
recognition based on image data. If you have any specific questions about the results or metrics
in deep learning for bird species identification.

FIG 6.2 : HOME PAGE

Department Of Computer Science, Mandya University, Mandya Page 40

 Birds Species Identification Using Deep Learning

FIG 6.2.1 : LOGIN PAGE

FIG 6.2.2 :OUTPUT PAGE

Department Of Computer Science, Mandya University, Mandya Page 41

 Birds Species Identification Using Deep Learning

FIG 6.2.3 : FINAL OUTPUT

Department Of Computer Science, Mandya University, Mandya Page 42

 Birds Species Identification Using Deep Learning

CHAPTER 7

SOFTWARE TESTING

7.1 INTRODUCTION

Testing is the significant interaction engaged with programming quality assurance (QA).
It is an iterative interaction. Here test information is arranged and is utilized to test the modules
exclusively. Framework testing ensures that all segments of the framework work appropriately as
a unit by really driving the framework to come up short.

The test causes ought to be arranged prior to testing starts. At that point as the testing
advances, testing shifts center trying to discover blunders in incorporated groups of modules and
in the whole framework. The way of thinking behind testing is to discover blunders. As a matter
of fact, testing is the domain of execution that is pointed toward guaranteeing that the framework
works really and proficiently before execution.

Testing is accomplished for every module. In the wake of testing every one of the
modules, the modules are incorporated and testing of the last framework is finished with the test
information, exceptionally intended to show that the framework will work effectively in the
entirety of its viewpoints conditions. The methodology level testing is made first. By giving ill-
advised data sources, the blunders that happened are noted and killed. The end-client instead of
the framework designer leads this test by most programming engineers as a cycle called the
"Alpha and Beta test" to uncover that solitary the end-client appears to be ready to discover.

This is the last advance in the framework life cycle. Here we carry out the tried blunder-
free framework into a genuine climate and roll out vital improvements, which runs in an online
design. Here framework upkeep is done each month or year dependent on organization
approaches and is checked for blunders like runtime mistakes, since a long time ago run
mistakes, and different systems of support like table confirmation and reports.

During the prerequisite examination and plan, the yield is a record that is typically text-
based and non-executable. After the coding stage, PC programs are accessible that can be

Department Of Computer Science, Mandya University, Mandya Page 43

 Birds Species Identification Using Deep Learning

executed for testing purposes. This infers that testing not just needs to uncover blunders
presented during coding, yet additionally mistakes presented during the past stages.

Fig. 7.1 The Testing process

7.1.1 UNIT TESTING

Unit testing confirmation endeavors on the littlest unit of programming plan, module.
This is known as "Module Testing". The modules are tried independently. This testing is done
during the programming stage itself. In these testing steps, every module is discovered to be
working acceptably with respect to the normal yield from the module.

7.1.2 INTEGRATION TESTING

Integration testing is a methodical strategy for developing tests to uncover blunders

related with the interface. In the venture, every one of the modules is joined and afterward the
whole developer is tried overall. In the combination testing step, the entire mistake revealed is
amended for the following testing steps.

Department Of Computer Science, Mandya University, Mandya Page 44

 Birds Species Identification Using Deep Learning

7.1.3 VALIDATION TESTING

To uncover functional errors, that is, to check whether functional characteristics confirm
to specification or not specified.

7.1.4 SYSTEM TESTING

When individual module testing finished, modules are gathered to proceed as a framework.
At that point, the hierarchical testing, which starts from the upper level to bring down level
module testing, must be done to check whether the whole framework is performing acceptably.
After unit and integration testing are over then the system as whole is tested.

7.1.5 ACCEPTANCE TESTING

At the point when the framework has no action issue with its precision, the framework goes
through a last acceptance test. This test affirms that the framework needs the first objective,
Objective, and necessities set up during the investigation. On the off chance that the framework
satisfies every one of the prerequisites, it is at last worthy and prepared for activity.

TEST PLAN

A product project test plan is an archive that depicts the destinations, scope approach, and
focal point of a product testing exertion. This cycle of setting up a test plan is a valuable method
to thoroughly consider the endeavors expected to approve the agreeableness of a product item.

TEST PLANS USED IN UNIT TESTING

Each module is tested for correctness whether it is meeting all the expected results.
Condition loops in the code are properly terminated so that they don’t enter into an infinite loop.
Proper validations are done so as to avoid any errors related to data entry from user.

SYSTEM TESTING

System Testing alludes to the way of looking at a product program utility fundamentally
dependent on what its particular says its conduct should be. In remarkable, we will develop

Department Of Computer Science, Mandya University, Mandya Page 45

 Birds Species Identification Using Deep Learning

check cases put together totally absolutely with respect to the detail of this current gadget's
conduct, without seeing an execution of this framework.

7.2 Unit Test Cases

The unit test cases help ensure that the deep learning model is functioning correctly and
accurately identifying bird species. By testing specific aspects of the model, you can catch errors
or bugs early on and improve the overall performance of the system.

Testing Scenario Expected result Result

Test Case Number
TC – 01 Uploading wrong Alert ”File extension Pass
data file not supported ”
TC – 02 Uploading without Alert ”Choose a file” Pass
choosing any file
TC – 03 View grayscaled Show correct Pass
image grayscaled image
TC – 04 View threshold Show correct Pass
image threshold image of
the bird
TC – 05 View segmented Show correct Pass
image segmented image of
bird
TC - 06 View Feature Show correct Pass
Extracted image featured extracted
image of bird
TC - 07 View Output Show correct bird Pass
species name

Table .7.2 Test case

Department Of Computer Science, Mandya University, Mandya Page 46

 Birds Species Identification Using Deep Learning

CHAPTER 8

CONCLUSION AND FUTURE ENHANCEMENT

In this work, methodology of classification of bird’s classification based on CNN is

proposed. This system tells us, which bird sample belong to which class in already stored dataset.
We aim to develop a quick, automatic and accurate system for the bird’s classification. Image
processing techniques are used for extracting the color parameter like RBG, HSV, CMYK, YIQ
that are having some significance and effect on the classification of birds. Extracted features of
birds based on the color and stored in database and apply the SVM method for classification.
First, we do the pre-processing stages and extract the components value of the model. In other
words, the results of this project can be used as a specification of bird’s features that must be
considered for bird’s classification.

In our project future work is we have worked on the limited feature which we have
concentrated only on color moment feature. The goal of the project is still far. Next step in the
future work are different classifier will be tested based on our dataset and result will be recorded.
Comprehensive between two or more classifiers will be more accurate.so that as been a
challenging task. Due to the number of factors like image condition, illumination and image
background, most of methodology and algorithms are unable to provide a solution for those
problems. A consideration attention is required to develop efficient method to overcome these
problems. Here we collect the standard database so in future collect the data from itself and do
the segmentation process. In future can remove the background of the image properly and
segment the image according to the user need. To get more accurate different methodology for
classification and feature extraction should be applied. We can develop a android application for
a better utilization.

Department Of Computer Science, Mandya University, Mandya Page 47

 Birds Species Identification Using Deep Learning

REFERENCES

[1] B. Vivekananda, "Speedy Image Crowd Counting by Light Weight Convolutional Neural
Network", Journal of Innovative Image Processing, vol. 3, no. 3, pp. 208-222, 2021.

[2] J. Samuel. Manoharan, "Study of Variants of Extreme Learning Machine (ELM) Brands and
its Performance Measure on Classification Algorithm", Journal of Soft Computing Paradigm
(JSCP), vol. 3, no. 02, pp. 83-95, 2021.

[3] A. M. Chandu B, "Automated Bird Species Identification using Audio", 2020 International
Conference on Artificial Intelligence and Signal Processing (AISP), 2020.

[4] R. P. Aarti Madhavi, "Deep Learning Based Audio Classifier for Bird Species", IJSDR,
2018.

[5] A. Incze, H.-B. Jancso, Z. Szilagyi, A. Farkas and C. Sulyok, "Bird Sound Recognition Using
Convolutional Neural Network", IEEE 16th International Symposium on Intelligent Systems and
Informatics, pp. 000295-000300, 2018.

[6] Atanbori, J., Duan, W., Murray, J., Appiah, K., & Dickinson, P. (2019). Automatic
classification of flying bird species using computer vision techniques. Pattern Recognition
Letters, 81, 53-62.

[7] Abdelkhalak, Bahri & Hamid, Zouaki. (2019). Content-Based Bird Retrieval using Shape
context, Color moments and Bag of Features.

[8] Alter, Anne Louise and Karen Wang. “An Exploration of Computer Vision Techniques for
Bird Species Classification.” (2018).

[9] Marini, Andreia & Facon, Jacques & Koerich, Alessandro. (2019). Bird Species
Classification Based on Color Features. Proceedings - 2013 IEEE International Conference on
Systems, Man, and Cybernetics, SMC 2013. 4336-4341. 10.1109/SMC.2013.740.

[10] Mehyadin, Aska & Mohsin Abdulazeez, Adnan & Hasan, Dathar & Saeed, Jwan. (2021).
Birds Sound Classification Based on Machine Learning Algorithms. Asian Journal of Research
in Computer Science. 1-11. 10.9734/AJRCOS/2021/v9i430227.

Department Of Computer Science, Mandya University, Mandya Page 48

 Birds Species Identification Using Deep Learning

[11] Aggarwal and S. Sehgal, "Classification of Bird Species using Audio processing and Deep
Neural Network," 2022 Third International Conference on Intelligent Computing
Instrumentation and Control Technologies (ICICICT), Kannur, India, 2022, pp. 138-143, doi:
10.1109/ICICICT54557.2022.9917735.

[12] Yang, Y. Jiang and Y. Xu, "Design of Bird Sound Recognition Model Based on
Lightweight," in IEEE Access, vol. 10, pp. 85189-85198, 2022, doi:
10.1109/ACCESS.2022.3198104.

[13] Jadhav, V. Patil and D. Parasar, "Machine Learning Approach to Classify Birds on the Basis
of Their Sound," 2020 International Conference on Inventive Computation Technologies
(ICICT), Coimbatore, India, 2020, pp. 69-73, doi: 10.1109/ICICT48043.2020.9112506.

[14] H. A. Jasim, S. R. Ahmed, A. A. Ibrahim and A. D. Duru, "Classify Bird Species Audio by
Augment Convolutional Neural Network," 2022 International Congress on Human-Computer
Interaction, Optimization and Robotic Applications (HORA), Ankara, Turkey, 2022, pp. 1-6,
doi: 10.1109/HORA55278.2022.9799968.

[15] M. M. M. Sukri, U. Fadlilah, S. Saon, A. K. Mahamad, M. M. Som and A. Sidek, "Bird

Sound Identification based on Artificial Neural Network," 2020 IEEE Student Conference on
Research and Development (SCOReD), Batu Pahat, Malaysia, 2020, pp. 342-345, doi:
10.1109/SCOReD50371.2020.9250746.

Department Of Computer Science, Mandya University, Mandya Page 49