Scribd
Upload
38 views

MINI PROJECT PPT

The project focuses on developing a machine learning system to detect oil spills in oceans using satellite images, aiming to automate detection and improve response times. It addresses the limitations of traditional methods, such as slow detection and limited coverage, by integrating advanced data analytics and remote sensing techniques. The system has shown promising results in distinguishing oil spills from similar features, highlighting its potential for environmental monitoring and protection.

Uploaded by

r77738736
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
38 views

MINI PROJECT PPT

The project focuses on developing a machine learning system to detect oil spills in oceans using satellite images, aiming to automate detection and improve response times. It addresses the limitations of traditional methods, such as slow detection and limited coverage, by integrating advanced data analytics and remote sensing techniques. The system has shown promising results in distinguishing oil spills from similar features, highlighting its potential for environmental monitoring and protection.

Uploaded by

r77738736
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 12

Project

Presentation

“OIL SPILL DETECTION”

MITU22BTCS0396:- Krushnal Patil


MITU22BTCS0143:- Archana Kumar
MITU22BTCS0473:- Neha Hambir
MITU22BTCS0958:- Vanshaj Saraf

Guided by:- Prof. Nilesh Kulal

Department of Computer Science & Engineering, MITSoC, Loni Kalbhor


Outline

● Introduction
● Problem Statement
● Objectives
● Concepts & Methods
● Literature Review
● Identification of gaps & scope of work
● Implementation
● Reference
● Q&A

Department of Computer Science & Engineering, MITSoE, Loni Kalbhor


2
Introduction
This project focuses on creating a system that uses satellite images and machine learning to detect oil spills in
the ocean. Oil spills can have serious negative effects on marine life and coastal communities, and traditional
methods of spotting them are often slow and require a lot of manual effort. Our approach aims to streamline
the process by automating detection, which could help environmental agencies respond more quickly and
effectively to these accidents. In this project, we explain how we harness remote sensing data, develop and
train our machine learning model, and test the system’s accuracy in distinguishing oil spills from other similar
features on the water.

Department of Computer Science & Engineering, MITSoE, Loni Kalbhor


3
Problem Statement
Oil spills pose a significant environmental threat to marine ecosystems, affecting
aquatic life and coastal economies. Early detection and timely response are crucial to
mitigating their impact. This project aims to develop a machine learning model that
classifies satellite images of water bodies as an oil spill or no spill.

Environmental Threat: Oil spills pose a serious threat to marine ecosystems, affecting
wildlife, water quality, and coastal communities, necessitating timely detection and
mitigation.
Delayed Detection: Traditional oil spill detection methods are often slow, labor-
intensive, and dependent on manual observation or limited satellite passes, leading to
delayed response times.
Limited Coverage: Current monitoring systems may lack real-time capability or have
limited spatial coverage, making it difficult to detect smaller or remote spills effectively.
Objectives
Integration with Response Systems: There is a lack of integration between
detection systems and emergency response frameworks, delaying containment and
cleanup operations once a spill is identified.
The objective of this project is to develop an automated and accurate oil spill detection
system using remote sensing data and machine learning techniques. It aims to enhance
detection accuracy, reduce false positives, and support timely response efforts to
minimize environmental impact and protect marine and coastal ecosystems.

Department of Computer Science & Engineering, MITSoE, Loni Kalbhor


4
Concepts and Methods
1) Satellite Imagery: Captures high resolution images of the
ocean surface to detect and monitor oil spill patterns and spread.

2) Remote Sensing: Utilizes various sensors on satellites to


analyze the oil's spectral signature and differentiate it from
water.

3) Data Analytics: Applies algorithms and machine learning to


interpret oil spill data and satellite images, improving spill
detection accuracy.

4) Aerial Surveillance (Drones and Aircraft):


Captures high-resolution images or videos to visually detect and
monitor oil spills, especially in remote areas.

Department of Computer Science & Engineering, MITSoE, Loni Kalbhor


5
Literature Survey
Sr. no Literature Title Author Findings

1. The Application of Satellite Image Analysis in Oil Paweł Tysiąc, This study explores the use of satellite
Spill Detection Tatiana Strelets, imagery for detecting oil spills, emphasizing
Weronika Tuszyńska the effectiveness of synthetic aperture radar
(SAR) data in identifying oil slicks under
various environmental conditions.

2. Hyperspectral Remote Sensing Benchmark Puhong Duan, The paper introduces an unsupervised method
Database for Oil Spill Detection with an Isolation Xudong Kang, for oil spill detection using hyperspectral
Forest-Guided Unsupervised Detector Pedram Ghamisi imagery, by employing an Isolation Forest
algorithm. The approach effectively
distinguishes oil spills from seawater without
the need for extensive labeled datasets.

3. Variable-Complexity Machine Learning Models Mohammad Danesh- Developed and compared machine learning
for Large-Scale Oil Spill Detection: The Case of Yazdi. models—Support Vector Machine (SVM),
Persian Gulf Sahand Najafizadegan Random Forest (RF), and Convolutional Neural
Network (CNN), Found that CNN outperformed
RF and SVM, achieving a testing accuracy of
95.8%, indicating its robustness for large-scale
oil spill detection.

Department of Computer Science & Engineering, MITSoE, Loni Kalbhor


6
Identification of gaps & scope of work
Identification of Gaps
Limited Dataset Availability: Lack of large, labeled datasets for training and evaluating machine learning models
specific to oil spills.
Detection Accuracy: Existing models may struggle with distinguishing oil spills from similar-looking features (e.g.,
algal blooms, ship waves).
Resolution Constraints: Many satellite images have medium to low resolution, reducing the ability to detect small-
scale spills.
Environmental Variability: Changing sea conditions (waves, wind, lighting) affect the reliability of detection
algorithms.

Scope of Work
Data Collection: Gather and preprocess satellite (SAR/optical) or drone-based images of oil spills and clean sea
surfaces.
Model Development: Design and train a machine learning or deep learning model capable of accurately identifying
oil spills.
Evaluation: Test model accuracy using performance metrics like precision, recall, F1-score, and confusion matrix.
Comparative Analysis: Compare different algorithms (e.g., SVM, CNN, Random Forest) to identify the most suitable
method.
Scalability: Investigate the model’s ability to be deployed at scale for large geographic areas using satellite data.

Department of Computer Science & Engineering, MITSoE, Loni Kalbhor


7
Implementation

Department of Computer Science & Engineering, MITSoE, Loni Kalbhor


8
Results
1) Economic Benefits: Cuts costs, boosts growth, and creates jobs.

2) Environmental Benefits: Reduces impact, promotes sustainability, and saves resources.

3) Technological: Improves efficiency, performance, innovation and scalability.

4) Robust Against Look-Alikes: Successfully distinguished between oil spills and similar features
like algae, ship wakes, and low-wind slicks in most test cases.

Department of Computer Science & Engineering, MITSoE, Loni Kalbhor


9
Conclusion and Future Work
To wrap it up, this project showed that using satellite images and machine learning
can be a smart and reliable way to detect oil spills in the ocean. The system we
built was pretty accurate and could tell the difference between actual oil spills and
other things that look similar, like algae or ship trails. It also cuts down on the time
and effort needed to spot spills manually. Overall, this kind of technology can
really help us keep a closer eye on our oceans and respond more quickly to
environmental accidents.

Department of Computer Science & Engineering, MITSoE, Loni Kalbhor


10
References
[1) Ngoc An Bui, Youngon Oh, and Impyeong Lee. "Oil spill detection and classification through deep learning and
tailored data augmentation." International Journal of Applied Earth Observation and Geoinformation.
2) Xiaorui Dong, Jiansheng Li, Bing Li, Yueqin Jin, and Shufeng Miao. "Marine oil spill detection from low-quality
SAR remote sensing images." Journal of Marine Science and Engineering.
3) Thomas De Kerf, Jona Gladines, Seppe Sels, and Steve Vanlanduit. "Oil Spill Detection Using Machine Learning
and Infrared Imaging." Remote Sensing.
4) Najafizadegan,S.,&Danesh-Yazdi,M.(2023) Variable-complexity machine learning models for large-scale oil spill
detection: The case of Persian Gulf.Marine Pollution Bulletin, 194, 115259.
5) Satellite Datasets (https://dataspace.copernicus.eu/).
6) SNAP tool for processing SAR datasets (https://step.esa.int/main).

Department of Computer Science & Engineering, MITSoE, Loni Kalbhor


11
Thank You

Department of Computer Science & Engineering, MITSoE, Loni Kalbhor


12

You might also like