1. Repositório Aberto da Universidade do Porto
2. FEUP - Faculdade de Engenharia
3. FEUP - Dissertação

Author(s):	João Filipe Alves Costa
Title:	Artificial Intelligence for Vineyard Trap Segmentation and Perspective Correction
Issue Date:	2021-07-21
Description:	This master thesis topic emerges in the EyesOnTraps project, which aims to develop a mobile solution for plague prevention, supporting automated insect detection in traps and treatment recommendation for detected plagues, in order to lessen human error and improve analysis. The handheld image acquisition process in open field is quite challenging, not only due to the heterogeneity and variability of light exposure, but also because the trap images need to adequate for further analysis (e.g. trap clearly visible and in an acceptable perspective). The incorporation of Artificial Intelligence (AI) into mobile applications to make the "handheld" image capture process simpler and more effective has already showed promising results in the health area. However, its potential in the viticulture sector still relies unexplored. This thesis aims to explore AI-based approaches to improve handheld image acquisition of vineyard traps via smartphones. In particular, innovative ways to segment these insect traps should be explored, for instance through automatic landmark detection using deep learning, which has the potential to deal with occlusions, folds, curved and/or partially broken corners of the trap in a more efficient way. Having the segmentation contours of the trap, a comparative study of the state-of-the-art approaches for perspective correction should be performed with the main goal of assessing the performance of each one.
Subject:	Engenharia electrotécnica, electrónica e informática Electrical engineering, Electronic engineering, Information engineering
Scientific areas:	Ciências da engenharia e tecnologias::Engenharia electrotécnica, electrónica e informática Engineering and technology::Electrical engineering, Electronic engineering, Information engineering
DOI:	10.34626/mk72-dn74
TID identifier:	202821102
URI:	https://hdl.handle.net/10216/135228
Document Type:	Dissertação
Rights:	openAccess
Appears in Collections:	FEUP - Dissertação

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