Publication
Vineyard Terrace Segmentation in the Douro Region Based on Satellite Imagery
| datacite.subject.fos | Engenharia e Tecnologia::Outras Engenharias e Tecnologias | pt_PT |
| dc.contributor.advisor | Martins, José Maria Gouveia | |
| dc.contributor.advisor | Oliveira, Ana Sofia Cartaxo Pinto | |
| dc.contributor.author | Lopes, Pedro António Sousa Pedrosa | |
| dc.date.accessioned | 2024-03-22T16:59:10Z | |
| dc.date.available | 2024-03-22T16:59:10Z | |
| dc.date.issued | 2023-11-15 | |
| dc.description.abstract | The Alto Douro Wine region holds the distinction of being a UNESCO World Heritage Site, known for its traditional vineyard terraces that contribute to its cultural significance. These terraces, engineered to support vine cultivation on the challenging slopes of the Douro valley, were affected by the Phylloxera pest outbreak in the 19th century, resulting in terrace reconstructions for disease control. Preserving this cultural landscape requires periodic evaluations of the terraces, but current manual field assessments are time-consuming, costly, and prone to errors, leading to infrequent updates. To address these challenges, this dissertation studies alternative approaches using multispectral and SAR satellite imagery, and machine learning to detect and identify vineyards within the terraces, aiming to reduce costs and increase assessment frequency. The study begins with a review of remote sensing and satellite imaging technologies, followed by a literature review on similar applications and techniques. Data acquisition details are provided, and three segmentation methodologies are explored: band indices, traditional machine learning (support vector machines and random forests) and deep learning (convolutional neural networks). The deep learning approach, particularly the modified DeepLabV3 model with the ResNet-101 backbone yields the most promising results, despite generalization limitations. Combining the segmented vineyard mask with a slope mask derived from SAR altimetry data increases confidence in identifying vineyards within terraces, offering rough estimations on possible locations of vineyard terraces in the Douro region. In conclusion, this study presents an alternative and cost-effective approach to preserve the heritage landscape of the Alto Douro Wine region. By leveraging satellite imagery and machine learning, it offers a practical and preliminary means for periodic evaluations, supporting the sustainable conservation of this culturally significant region. | pt_PT |
| dc.identifier.tid | 203561422 | pt_PT |
| dc.identifier.uri | http://hdl.handle.net/10400.8/9569 | |
| dc.language.iso | eng | pt_PT |
| dc.subject | Machine learning | pt_PT |
| dc.subject | Deep learning | pt_PT |
| dc.subject | Remote sensing | pt_PT |
| dc.subject | Earth Observation | pt_PT |
| dc.subject | Alto Douro Vineyard | pt_PT |
| dc.title | Vineyard Terrace Segmentation in the Douro Region Based on Satellite Imagery | pt_PT |
| dc.type | master thesis | |
| dspace.entity.type | Publication | |
| rcaap.rights | openAccess | pt_PT |
| rcaap.type | masterThesis | pt_PT |
| thesis.degree.name | Mestrado em Ciência de Dados | pt_PT |
Files
Original bundle
1 - 1 of 1
Loading...
- Name:
- thesis_vineyard_terrace_segmentation_updated_correções_formais.pdf
- Size:
- 5.72 MB
- Format:
- Adobe Portable Document Format
- Description:
License bundle
1 - 1 of 1
No Thumbnail Available
- Name:
- license.txt
- Size:
- 1.32 KB
- Format:
- Item-specific license agreed upon to submission
- Description: