Please use this identifier to cite or link to this item: http://hdl.handle.net/10609/152264
Title: Ús de deep learning i imatges de drons per a la detecció de tortugues marines en entorns naturals
Author: Mahamud El Jalil, Hafdal-la
Director: Bas Pujols, Bernat
Tutor: Acedo Nadal, Susana
Abstract: Extraction and analysis of images using computer vision techniques has become more popular in recent years. It has applications in scientific, social, economic, and military fields. At the same time, the use of drones or unmanned aerial vehicles (UAVs) has also become more widespread in recent years due to increased accessibility and improvements in their capabilities, allowing the capture of high-quality images in various contexts. This work aims to combine these two technologies with an environmental purpose: to contribute to the study and protection of sea turtles in areas where they are at risk of dissapearing. Human monitoring in these areas is costly and time-consuming. The use of deep learning algorithms requires less time and other resources and thus allows for resources to be spent on higher-quality tasks by limiting human intervention up to the point of final verification. The objective of this work is to achieve highly accurate results, minimize false negatives, and ensure that the created model is reliable and adds value to the organizations involved in the protection of sea turtles by allowing for more efficient and scalable monitoring. This would facilitate work across a larger number of regions or a higher volume of recordings. The YOLO (You Only Look Once) algorithm was used to detect sea turtles. To evaluate the performance of the models, metrics of precision, sensitivity, and IoU were used. The result was a model capable of detecting turtles with a precision of 83.1% and a sensitivity of 84.8%. However, the model struggles to detect turtles in more complex settings and would require further improvements to detect turtles in a wider range of settings.
Keywords: Deep learning
YOLO
Computer vision
Automatic classification
Accuracy
Dron/es
Sea turtles
Biodiversity
Animal protection
Document type: info:eu-repo/semantics/masterThesis
Issue Date: 9-Jan-2025
Publication license: http://creativecommons.org/licenses/by-nc-nd/3.0/es/  
Appears in Collections:Bachelor thesis, research projects, etc.

Files in This Item:
File Description SizeFormat 
hmahamudeljalilTFM0125.pdfTFM12,35 MBAdobe PDFThumbnail
View/Open
Share:
Export:
View statistics

Items in repository are protected by copyright, with all rights reserved, unless otherwise indicated.