CHAPTER 3
REQUIREMENTS AND ANALYSIS
3.1 PROBLEM DEFINITION:
In recent years, the problem of waste management has become increasingly severe due to
rapid urbanization and population growth. One of the major challenges in this field is the manual
segregation of garbage, which is not only time-consuming and labor-intensive but also prone to
errors. Improper garbage classification can lead to ineffective recycling processes, increased
landfill usage, and harmful environmental impacts.
The current systems of waste segregation largely rely on human intervention or traditional rule-
based methods, which lack accuracy, adaptability, and scalability. In this context, the need arises
for an Automated and Intelligent garbage classification system that can accurately categorize
different types of waste such as biodegradable, recyclable, and hazardous materials.
To address this, the project proposes the use of Transfer learning a machine learning technique
where a pre-trained deep learning model is adapted for a new but related task. By leveraging
powerful models like EfficientNetV2B2 , the system can recognize and classify waste images with
high accuracy even with limited labeled data.
The goal is to build a system that:
Takes an image of a waste item as input,
Automatically classifies it into the correct category (e.g., Cardboard, plastic, metal, paper),
Helps streamline the waste segregation process for smart cities, recycling centers, and
public waste bins.
This project aims to bridge the gap between conventional waste management techniques and
intelligent automation by using the capabilities of deep learning and transfer learning.
3.2 REQUIREMENT SPECIFICATION:
3.2.1 FUNCTIONAL REQUIREMENTS:
The system is designed to perform several key functions that enable accurate and efficient
garbage classification. First and foremost, it should allow users to upload or capture images of
garbage items through a user interface. Once an image is provided, the system must process the
input, extract relevant features, and classify the image into appropriate categories such as
Cardboard, paper, metal, trash, plastic waste. The classification process will be powered by a deep
learning model trained using transfer learning. Additionally, the system should display the
prediction result along with the confidence level and possibly suggestions on how to dispose of or
recycle the item. A database may be used to store image data, prediction results, and user feedback
to improve the model over time. Lastly, the system should provide a mechanism for users to report
incorrect classifications, which can help in refining the model through future updates.
3.2.2 NON FUNCTIONAL REQUIREMENTS:
The system must be reliable and provide accurate predictions with minimal errors,
maintaining at least a 90% classification accuracy during real-world testing. It should be
responsive and able to classify garbage items in real-time or within a few seconds to ensure a
smooth user experience. The solution should be scalable, allowing it to handle a growing number
of users and images without performance degradation. Portability is also important—the system
should run effectively on both desktop and mobile devices. Furthermore, the user interface must
be intuitive and easy to use, even for individuals with limited technical knowledge. Security and
data privacy must be ensured, especially if the system collects user images or stores data in the
cloud. Finally, the system should be maintainable and well-documented to allow for future updates
or improvements by other developers.
3.3 PLANNING AND SCHEDULING:
Effective planning and scheduling are essential to ensure the successful and timely completion
of the project. The entire project has been divided into several key phases: Planning, Requirements
Gathering, Design, Frontend Development, Backend Development, Testing, and Bug Fixing.
The following table outlines the planned schedule for each phase:
Table 3.1 Project Plan and Schedule
Phase Start Date End Date Duration
Planning 10-07-2025 15-07-2025 6 days
Requirements
16-07-2025 22-07-2025 7 days
Gathering
Design 23-07-2025 31-07-2025 9 days
Development
01-08-2025 14-08-2025 14 days
(Frontend)
Development
15-08-2025 31-08-2025 17 days
(Backend)
Testing 01-09-2025 15-09-2025 15 days
Bug Fixing 16-09-2025 30-09-2025 15 days
Figure 3.1 Gantt Chart
3.4 SOFTWARE AND HARDWARE REQUIREMENTS:
3.4.1 SOFTWARE REQUIREMENTS:
Environment : Anaconda Navigator and Jupyter Notebook
Frontend : Gradio / Streamlit
Backend : Python with TensorFlow and Keras
Libraries and Tools : NumPy, Pandas, Matplotlib, Seaborn, EfficientNetV2B2, OpenCV
3.4.2 HARDWARE REQUIREMENTS:
RAM : 8 GB
Processor : Intel Core i5
Operating System : Windows 10 (64-bit)
Storage : SSD with 250 GB
Processor Speed : Minimum 2.0 GHz
3.5 CONCEPTUAL MODELS:
The conceptual model represents the high-level structure and flow of the garbage classification
system. It outlines how different components such as input data, machine learning models, and
output classes interact with each other. The system works by receiving an image of a garbage item,
processing it using a transfer learning model, and classifying it into one of the predefined
categories.
3.5.1 ER DIAGRAM:
Figure 3.2 ER Diagram
