CSE-MINOR PROJECT-1 Review 2 RECYCLABLE SOLID WASTE DETECTION USING DEEP LEARNING.pdf

The project aims to automate solid waste classification into recyclable and non-recyclable categories using Deep Learning.

The purpose is to assist waste management authorities and recycling units by reducing human error in waste segregation.

The method involves using Convolutional Neural Networks (CNNs) trained on image datasets of various waste materials.

The system identifies recyclable materials such as paper, metal, plastic, and glass in real time.

Results show improved accuracy and speed over manual sorting methods.

Conclusion: The system provides an AI-driven, sustainable approach for waste segregation and environmental protection.

1. DEPARTMENT OF COMPUTER SCIENCE & ENGINEERING
SCHOOL OF COMPUTING
10214CS601 MINOR PROJECT -1
SUMMER SEMESTER(2025 -2026)
REVIEW-II


“RECYCLABLE SOLID WASTE DETECTION USING DEEP LEARNING”
SUPERVISED BY
1DEPARTMENT OF COMPUTER SCIENCE & ENGINEERING / RECYCLABLE SOLID WASTE DETECTION USING DEEP LEARNINGOctober 24, 2025
PRESENTED BY
1.Oduru Saichaithanya (VTU26296) (23UECS0410)
2.Pathuri SujithReddy (VTU26220) (23UECS0440)
3.Papareddy Rahul (VTU26295) (23UECS0430)
Dr.Rajeswari RajeshImmanuel

2. October 24, 2025 2
OVERVIEW
❑ABSTRACT
❑OBJECTIVE
❑INTRODUCTION
❑LITERATURE REVIEW (SOFT COPY OF PAPERS TO BE LINKED AS HYPERLINK)
❑DESIGN AND METHODOLOGIES
❑IMPLEMENTATION
❑TESTING
❑INPUT AND OUTPUT
❑INCLUDE DEMO VIDEO-1 (Till REVEW-1)
❑INCLUDE DEMO VIDEO-2(Complete Implementation of Project)
❑CONCLUSION
❑WEB REFERENCES LINK (TILL REVIEW DATE ALL LINKS TO BE INCLUDED DAY WISE)
❑PLAGIARISM REPORT OF PPT
❑REFERENCES
DEPARTMENT OF COMPUTER SCIENCE & ENGINEERING / RECYCLABLE SOLID WASTE DETECTION USING DEEP LEARNING

3. October 24, 2025 3
ABSTRACT
➢The project aims to automate solid waste classification into recyclable and non-recyclable
categories using Deep Learning.
➢The purpose is to assist waste management authorities and recycling units by reducing
human error in waste segregation.
➢The method involves using Convolutional Neural Networks (CNNs) trained on image
datasets of various waste materials.
➢The system identifies recyclable materials such as paper, metal, plastic, and glass in real
time.
➢Results show improved accuracy and speed over manual sorting methods.
➢Conclusion: The system provides an AI-driven, sustainable approach for waste
segregation and environmental protection.
DEPARTMENT OF COMPUTER SCIENCE & ENGINEERING / RECYCLABLE SOLID WASTE DETECTION USING DEEP LEARNING

4. October 24, 2025 4
OBJECTIVES
Aim:
➢To design and implement an intelligent system that detects and classifies
recyclable solid waste using deep learning techniques.
Scope:
➢Applicable for smart cities, recycling plants, and waste segregation
units.
➢Can be deployed on edge devices or IoT systems for real-time
classification.
➢Promotes automation and sustainability in waste management.
DEPARTMENT OF COMPUTER SCIENCE & ENGINEERING / RECYCLABLE SOLID WASTE DETECTION USING DEEP LEARNING

5. October 24, 2025 5
TIMELINE OF THE PROJECT
DEPARTMENT OF COMPUTER SCIENCE & ENGINEERING / RECYCLABLE SOLID WASTE DETECTION USING DEEP LEARNING

6. October 24, 2025 6
INTRODUCTION
➢Improper waste segregation leads to environmental pollution and recycling inefficiency.
➢Manual sorting is time-consuming, costly, and prone to errors.
➢Deep learning can automate image-based waste classification efficiently.
➢The project leverages computer vision and CNNs to distinguish recyclable vs non-recyclable
waste.
➢This system supports sustainable waste management in urban areas.
DEPARTMENT OF COMPUTER SCIENCE & ENGINEERING / RECYCLABLE SOLID WASTE DETECTION USING DEEP LEARNING

7. October 24, 2025 7
LITERATURE REVIEW
Sl.NoAuthor’s Name Paper name and publication
details
Year of
publication
Main content of the
paper
1 Zhang et al. “Waste Classification Using CNN and
Transfer Learning,” IEEE Access
2022 Proposed CNN-based transfer
learning model for accurate
waste image classification.
2 Patel & Singh “Deep Learning Models for Smart Waste
Management,” IEEE IoT Journal
2023 Developed deep learning
approach integrating IoT for
smart waste collection and
monitoring.
3 Li et al. “YOLO-Based Object Detection for Solid
Waste Sorting,” IEEE ICMLA
2021 Implemented YOLO
algorithm for real-time waste
detection and classification in
recycling systems.
4 Wang et al. “Plastic Waste Recognition Using ResNet
Architecture,” IEEE Transactions on
Consumer Electronics
2024 Applied ResNet model for
identifying and classifying
different types of plastic
waste.
DEPARTMENT OF COMPUTER SCIENCE & ENGINEERING / RECYCLABLE SOLID WASTE DETECTION USING DEEP LEARNING

8. October 24, 2025 8
5 Kumar et al. “Real-Time Waste Detection Using
Edge AI Devices,” IEEE Sensors
Journal
2023Focused on deploying waste
detection models on edge devices
for faster, low-power processing
6 Ahmed & Roy “AI-Driven Waste Segregation Using
CNN,” IEEE Access
2022Introduced CNN-based classifier
achieving high accuracy in
segregating recyclable waste.
7 Chen et al. “Vision-Based Smart Bin for Waste
Management Using Deep Learning,”
IEEE ICME
2023Designed smart bins with embedded
deep learning models for automatic
waste categorization.
8 Gupta & Das “Optimization of Deep Models for
Sustainable Waste Classification,” IEEE
Transactions on AI
2025Proposed optimized CNN model to
improve waste classification
efficiency and reduce computation
time.
DEPARTMENT OF COMPUTER SCIENCE & ENGINEERING / RECYCLABLE SOLID WASTE DETECTION USING DEEP LEARNING

9. October 24, 2025 9
DESIGN AND METHODOLOGIES
DEPARTMENT OF COMPUTER SCIENCE & ENGINEERING / RECYCLABLE SOLID WASTE DETECTION USING DEEP LEARNING
Outline the methodologies that showcase how your project differs or improves:
➢MODULE 1: Transfer Learning
➢MODULE 2:VGG16

10. October 24, 2025 10DEPARTMENT OF COMPUTER SCIENCE & ENGINEERING / RECYCLABLE SOLID WASTE DETECTION USING DEEP LEARNING
Module -1 Transfer learning
Step:1 Collection of data
➢Surveys are a great way to collect data from a large number of people. They can be conducted in person,
over the phone, or online.
➢Experiments are used to test a hypothesis by manipulating one or more variables and observing the results.
➢Observational studies involve observing people or objects and recording their behavior or characteristics.
➢The data collection process can be time-consuming and challenging, but it is essential for any data
analysis project. By carefully collecting and preparing data, researchers can ensure that their results
are accurate and reliable.

11. October 24, 2025 11DEPARTMENT OF COMPUTER SCIENCE & ENGINEERING / RECYCLABLE SOLID WASTE DETECTION USING DEEP LEARNING
Step2:Processingthedata
Datanormalization:Thisinvolvesscalingthedatatoaspecificrange,typicallybetween0and1or-1and1.Normalizationensures
thatallfeatureshaveasimilarscale,preventinganysinglefeaturefromdominatingthemodel'slearningprocess.
Datastandardization:Thisinvolvestransformingthedatatohaveameanofzeroandastandarddeviationofone.
StandardizationachievesasimilargoalasnormalizationbutisparticularlyusefulwhenthedatahasaGaussiandistribution.
Dataaugmentation:Thisinvolvesartificiallyincreasingthesizeanddiversityofthedatasetbygeneratingnewdatasamples
fromtheexistingones.Dataaugmentationcanbeparticularlyusefulwhendealingwithlimiteddataavailability.
Imageresize:ResizingallowsyoutomakeyourimagesmallerorlargerwithoutcuttinganythingoutResizingalterstheimage's
dimensions,whichtypicallyaffectsthefilesizeandimagequality.Themostcommonreasonforresizingphotosistoreducethe
sizeoflargefilestomakethemeasiertoemailorshareonline.

12. October 24, 2025 12DEPARTMENT OF COMPUTER SCIENCE & ENGINEERING / RECYCLABLE SOLID WASTE DETECTION USING DEEP LEARNING
Modules -2-VGG16

➢VGG16 is a convolutional neural network (CNN) architecture with a simple and straightforward design.
➢It consists of 16 layers, including 13 convolutional layers, 5 max pooling layers, and 3 fully connected layers.
➢The convolutional layers use small 3x3 filters and a stride of 1, allowing for a deeper network without increasing
computational complexity.
➢Max pooling layers are used to reduce the dimensionality of the feature maps, reducing computational costs
and introducing invariance to small translations

13. October 24, 2025 13
Step 3: Using algorithm’s name algorithm
DEPARTMENT OF COMPUTER SCIENCE & ENGINEERING / RECYCLABLE SOLID WASTE DETECTION USING DEEP LEARNING

14. October 24, 2025 14
Step 4: The output
DEPARTMENT OF COMPUTER SCIENCE & ENGINEERING / RECYCLABLE SOLID WASTE DETECTION USING DEEP LEARNING

15. October 24, 2025 15
IMPLEMENTATION
➢Architecture Diagram
➢Data –Flow Diagram
➢Use Case Diagram
➢Class Diagram
➢Activity Diagram
➢Sequence Diagram
DEPARTMENT OF COMPUTER SCIENCE & ENGINEERING / RECYCLABLE SOLID WASTE DETECTION USING DEEP LEARNING

16. October 24, 2025 16
Architecture Diagram
DEPARTMENT OF COMPUTER SCIENCE & ENGINEERING / RECYCLABLE SOLID WASTE DETECTION USING DEEP LEARNING

17. October 24, 2025 17
Data –Flow Diagram
DEPARTMENT OF COMPUTER SCIENCE & ENGINEERING / RECYCLABLE SOLID WASTE DETECTION USING DEEP LEARNING

18. October 24, 2025 18
Use Case Diagram
DEPARTMENT OF COMPUTER SCIENCE & ENGINEERING / RECYCLABLE SOLID WASTE DETECTION USING DEEP LEARNING

19. October 24, 2025 DEPARTMENT OF COMPUTER SCIENCE & ENGINEERING / PROJECT TITLE 19
Class Diagram

20. October 24, 2025 20
Activity Diagram
DEPARTMENT OF COMPUTER SCIENCE & ENGINEERING / RECYCLABLE SOLID WASTE DETECTION USING DEEP LEARNING

21. October 24, 2025 21
Sequence Diagram
DEPARTMENT OF COMPUTER SCIENCE & ENGINEERING / RECYCLABLE SOLID WASTE DETECTION USING DEEP LEARNING

22. October 24, 2025 22
➢UNIT TESTING
➢INTEGRATION TESTING
➢FUNCTIONAL TESTING
➢WHITE BOX TESTING
➢BLACK BOX TESTING
TESTING
DEPARTMENT OF COMPUTER SCIENCE & ENGINEERING / RECYCLABLE SOLID WASTE DETECTION USING DEEP LEARNING

23. October 24, 2025 23
UNIT TESTING
Purpose: To test individual components (functions/modules) of the system.
DEPARTMENT OF COMPUTER SCIENCE & ENGINEERING / RECYCLABLE SOLID WASTE DETECTION USING DEEP LEARNING
Component Test Objective
Image preprocessing Check if images are resized and normalized correctly
Model loader Verify CNN model loads without error
(recycle_model.h5)
Data generator Ensure dataset is read correctly from train/ and val/
folders
Prediction function Check if predicted class returns correct label
Example:
Test if model loads correctly
Test if a single image returns a class output

24. October 24, 2025 24DEPARTMENT OF COMPUTER SCIENCE & ENGINEERING / RECYCLABLE SOLID WASTE DETECTION USING DEEP LEARNING
INTEGRATION TESTING
Integration testing is performed after unit testing, and its main purpose is to verify that individual modules of the system work correctly
when combined together. Even if each unit works independently, the system may still fail when components interact with each other.
Integration testing ensures smooth data flow and communication across modules.
In our project “Recyclable Solid Waste Detection Using Deep Learning”, integration testing is required, because multiple modules
interact with each other during the detection process, such as:
Integrated Modules Test Scenario
Dataset + DataGenerator Ensure images from folders are converted into batches
DataGenerator + CNN Model Check if model accepts generated batches
CNN Model + Prediction Verify probability output passes to UI/console
Trained model + Camera/Local file Test real-time & offline detection
Example:
Testing if camera input is processed and passed to CNN model properly

25. October 24, 2025 25
SYSTEM TESTING
DEPARTMENT OF COMPUTER SCIENCE & ENGINEERING / RECYCLABLE SOLID WASTE DETECTION USING DEEP LEARNING
System testing validates the entire system as a whole based on requirements.
What was tested Objective
End-to-end workflow Verify entire flow: input → detection → output
Performance Accuracy of classification on test data
File handling Model file (.h5) loading successfully
Example:
User gives an image → Model predicts correct recyclable class → Displays output

26. October 24, 2025 26DEPARTMENT OF COMPUTER SCIENCE & ENGINEERING / RECYCLABLE SOLID WASTE DETECTION USING DEEP LEARNING
FUNCTIONAL TESTING
Ensures that all functional requirements mentioned in the SRS are met.
Function Tested For
Model Training Correct number of epochs and output
Waste Classification Returns correct recyclable category
Notification/Output Proper display of result on UI or console
Example:
Paper image → outputs “Paper”
Plastic bottle → outputs "Plastic"

27. October 24, 2025 27DEPARTMENT OF COMPUTER SCIENCE & ENGINEERING / RECYCLABLE SOLID WASTE DETECTION USING DEEP LEARNING
WHITE BOX TESTING
This tests the internal logic of the code.
Area What is verified
CNN layers Activation and filter counts
Preprocessing steps Normalization operations
Training loop Loss and accuracy computations
Example:
Checking intermediate tensor shapes while forwarding image

28. October 24, 2025 28DEPARTMENT OF COMPUTER SCIENCE & ENGINEERING / RECYCLABLE SOLID WASTE DETECTION USING DEEP LEARNING
BLACK BOX TESTING
Here, only inputs and expected outputs are tested without looking at internal code.
Input Expected Output
Image of cardboard “Cardboard” detected
Image of e-waste “E-waste” detected
Non-recyclable waste “Trash” detected
Example:
Unknown test image → Model must still classify into correct category

29. October 24, 2025 29
INPUT AND OUTPUT
DEPARTMENT OF COMPUTER SCIENCE & ENGINEERING / RECYCLABLE SOLID WASTE DETECTION USING DEEP LEARNING

30. October 24, 2025 30
SOURCE CODE
DEPARTMENT OF COMPUTER SCIENCE & ENGINEERING / RECYCLABLE SOLID WASTE DETECTION USING DEEP LEARNING
# train.py SOURCE CODE

31. October 24, 2025 31
SOURCE CODE
DEPARTMENT OF COMPUTER SCIENCE & ENGINEERING / RECYCLABLE SOLID WASTE DETECTION USING DEEP LEARNING
# camera_detect.py SOURCE CODE

32. October 24, 2025 32
OUTPUT
DEPARTMENT OF COMPUTER SCIENCE & ENGINEERING / RECYCLABLE SOLID WASTE DETECTION USING DEEP LEARNING

33. October 24, 2025 33
CONCLUSION
Aim of the Project
➢To develop an intelligent system that can automatically detect and classify recyclable solid waste using deep
learning.
➢To promote smart waste segregation for cleaner environments and improved recycling efficiency.
➢To help users identify different waste categories such as plastic, paper, glass, metal, cardboard, and e-waste using
image-based recognition.
Problem Statement
➢Improper waste segregation leads to environmental pollution and low recycling rates.
➢Manual sorting of waste is time-consuming, inaccurate, and unsafe for workers.
➢Lack of automated systems causes recyclable waste to get mixed with non-recyclable materials, reducing reuse
potential.
➢There is a need for a machine learning-based solution that can accurately identify recyclable waste categories in
real-time.
.
DEPARTMENT OF COMPUTER SCIENCE & ENGINEERING / RECYCLABLE SOLID WASTE DETECTION USING DEEP LEARNING

34. October 24, 2025 34
Plagiarism Report of PPT
Plagiarism should be less than 10%.
Eg,
DEPARTMENT OF COMPUTER SCIENCE & ENGINEERING / RECYCLABLE SOLID WASTE DETECTION USING DEEP LEARNING

35. October 24, 2025 35
Web references/video links
DEPARTMENT OF COMPUTER SCIENCE & ENGINEERING / RECYCLABLE SOLID WASTE DETECTION USING DEEP LEARNING
Video Tutorials & Demonstrations
1.Recyclable Waste Classifier using OpenCV and Python
A step-by-step guide to building a waste classifier using OpenCV and Python.
YouTube
2.A Real-time Trash Detection Web App
Demonstrates an app that helps users determine if an item is recyclable or belongs in the trash.
YouTube
3.DEEP LEARNING TECHNIQUES FOR GARBAGE CLASSIFICATION
Explores deep learning techniques specific to garbage classification and how AI models are trained to sort and identify recyclable materials.
YouTube
Research Papers & Articles
1.WasteInNet: Deep Learning Model for Real‐time Waste Detection
A study on a deep learning model designed for real-time waste detection, categorized by type.
ScienceDirect
2.Intelligent Waste Sorting for Urban Sustainability using Deep Learning
Discusses hierarchical classification of waste to increase recovery by assigning each waste flow to either recycling/reutilization or final
disposal options.
Nature

36. October 24, 2025 36DEPARTMENT OF COMPUTER SCIENCE & ENGINEERING / RECYCLABLE SOLID WASTE DETECTION USING DEEP LEARNING
GitHub Projects & Code Repositories
1.teamsmcorg/Waste-Classification-using-YOLOv8
A project that focuses on accurately classifying waste into six different types using the YOLOv8 model.
GitHub
3.Plastic Waste Classification Using Deep Learning: Insights from the WaDaBa Dataset
Explores the potential of deep learning, focusing on convolutional neural networks (CNNs) and object detection models like YOLO,
to classify plastic waste using the WaDaBa dataset.
arXiv
Web references/video links

37. October 24, 2025 37
REFERENCES
[1] Zhang, Y., Li, X., and Wang, H., “Deep Learning-Based Recyclable Waste Classification Using YOLO and CNN
Models,” in Proceedings of the 2023 International Conference on Artificial Intelligence and Environmental Sustainability
(AIES’23), Beijing, China, June 2023, pp. 112–118.
[2] Greyparrot, "Deep Learning and Waste Recognition," Greyparrot, July 30, 2020. [Online]. Available:
https://www.greyparrot.ai/resources/blog/deep-learning-and-waste-recognition. [Accessed: Oct. 21, 2025].
[3] Sivacki, N., "Enhancing trash classification in smart cities using federated deep learning," Scientific Reports, vol. 14, no.
1, Article 62003, 2024. [Online]. Available: https://www.nature.com/articles/s41598-024-62003-4. [Accessed: Oct. 21,
2025].
[4] Zhang, Y., et al., "Real-time intelligent garbage monitoring and efficient collection using Yolov8 and Yolov5 deep
learning models for environmental sustainability," Scientific Reports, vol. 15, Article 99885, 2025. [Online]. Available:
https://www.nature.com/articles/s41598-025-99885-x. [Accessed: Oct. 21, 2025].
[5] Li, X., et al., "Enhancing waste sorting and recycling efficiency: robust deep learning-based approach for classification
and detection," Neural Computing and Applications, 2024. [Online]. Available:
https://link.springer.com/article/10.1007/s00521-024-10855-2. [Accessed: Oct. 21, 2025].
DEPARTMENT OF COMPUTER SCIENCE & ENGINEERING / RECYCLABLE SOLID WASTE DETECTION USING DEEP LEARNING

38. October 24, 2025 38
REFERENCES
DEPARTMENT OF COMPUTER SCIENCE & ENGINEERING / RECYCLABLE SOLID WASTE DETECTION USING DEEP LEARNING
[6] Zhang, Y., et al., "Skip-YOLO: Domestic Garbage Detection Using Deep Learning Method in Complex Multi-scenes,"
International Journal of Computational Intelligence Systems, vol. 18, no. 3, Article 314, 2023. [Online]. Available:
https://link.springer.com/article/10.1007/s44196-023-00314-6. [Accessed: Oct. 21, 2025].
[7] Sivacki, N., "Deep Learning and Waste Recognition," Greyparrot, July 30, 2020. [Online]. Available:
https://www.greyparrot.ai/resources/blog/deep-learning-and-waste-recognition. [Accessed: Oct. 21, 2025].
[8] Sivacki, N., "Deep Learning and Waste Recognition," Greyparrot, July 30, 2020. [Online]. Available:
https://www.greyparrot.ai/resources/blog/deep-learning-and-waste-recognition. [Accessed: Oct. 21, 2025].
[9] Sivacki, N., "Deep Learning and Waste Recognition," Greyparrot, July 30, 2020. [Online]. Available:
https://www.greyparrot.ai/resources/blog/deep-learning-and-waste-recognition. [Accessed: Oct. 21, 2025].
[10] Sivacki, N., "Deep Learning and Waste Recognition," Greyparrot, July 30, 2020. [Online]. Available:
https://www.greyparrot.ai/resources/blog/deep-learning-and-waste-recognition. [Accessed: Oct. 21, 2025].

39. 39October 24, 2025 DEPARTMENT OF COMPUTER SCIENCE & ENGINEERING / RECYCLABLE SOLID WASTE DETECTION USING DEEP LEARNING