Bird identification with advanced deep learning techniques for biodiversity conservation

The main objective of this thesis is to contribute to the conservation of the Galapagos Islands through education. To achieve this, an application has been developed that can run on both web and mobile devices, with the latter being the project’s strong point. The thesis presents two key contributions: Firstly, we conducted an exhaustive search for Galapagos birds using three different sources to generate a training dataset, as none existed for our approach. These carefully obtained and labeled data are based on the guidance from Charles Darwin’s Foundation main page. Secondly, neural networks and knowledge transfer were employed to develop a bird classifier, contributing to its conservation. Neural networks were trained using 58 different species of birds, whether native, endemic, or visitors. Once a satisfactory performance was achieved in the bird species identification model, a web/mobile application was developed. A solution allowing hosting the application on a web server was used, enabling remote access for users from mobile devices at any location. The purpose is to export the trained model to the server, facilitating bird species identification through image capture. The application features an initial interface that categorizes users as children, tourists, or biologists, adapting the provided information accordingly. Users can report sightings through calls generated from their mobile phones, directly contacting offices, or interacting with guides during island exploration. These actions are crucial for identifying native, endemic, and invasive bird species, contributing significantly to its conservation by preventing the introduction of non-native species that could endanger the biodiversity of the Galapagos Islands, whether due to human activities or natural factors.

• Príomhchruthaitheoir: Farinango Romero, Ricardo Joseph (author)
• Formáid: bachelorThesis
• Teanga: eng
• Foilsithe / Cruthaithe: 2024
• Ábhair: Biodiversidad; Redes neuronales; Transferencia del conocimiento; Biodiversity; Neural networks; Knowledge transfer
• Rochtain ar líne: http://repositorio.yachaytech.edu.ec/handle/123456789/775