The work aims at discovering the potential and the efficiency of Unmanned Aerial Systems (UAS) and Machine Learning (ML) in agriculture scenario, focusing on crop management and agrochemicals distribution optimization in orchard and horticultural cropping systems. The dissertation includes a general introduction, three experimental chapters and a general conclusion. Chapter 2 illustrates an operational approach to estimate individual and aggregate vineyards’ canopy volume using the manual Tree-Row-Volume (TRV) and the remotely sensed Canopy Height Model (CHM) techniques, processed with MATLAB scripts, and validated through ArcGIS tools. The results confirm how the extensive use of TRV is recommended when supported by remote sensing, to better qualify errors and heterogeneities in field estimates. Chapter 3 presents the development of a grape bunch detector based on a deep convolutional neural network trained to work directly on the field in an uncontrolled environment. The presented results are promising since most of the bunches were correctly detected with a 91% mean average precision, not only on the GrapeCS-ML database used to train the system, but also on an internal dataset, confirming the portability to different scenarios. Chapter 4 reports artichoke plant deep learning-based detection and georeferencing as the first step for an on-the-fly UAS spraying system and uses the gathered information to crop development monitoring in a multi-temporal approach. The Feature Pyramid Network, trained and compared with the YOLOv5 network, showed a high detection level with an average F1 score of around 90%, and satisfactory off-line performances on the Nvidia Jetson Nano board. The multi-temporal approach influenced detection performances, with an inverse response of precision and recall metrics. The growing index trend showed a distinct value in October, peaking at the beginning of December as expected.

The work aims at discovering the potential and the efficiency of Unmanned Aerial Systems (UAS) and Machine Learning (ML) in agriculture scenario, focusing on crop management and agrochemicals distribution optimization in orchard and horticultural cropping systems. The dissertation includes a general introduction, three experimental chapters and a general conclusion. Chapter 2 illustrates an operational approach to estimate individual and aggregate vineyards’ canopy volume using the manual Tree-Row-Volume (TRV) and the remotely sensed Canopy Height Model (CHM) techniques, processed with MATLAB scripts, and validated through ArcGIS tools. The results confirm how the extensive use of TRV is recommended when supported by remote sensing, to better qualify errors and heterogeneities in field estimates. Chapter 3 presents the development of a grape bunch detector based on a deep convolutional neural network trained to work directly on the field in an uncontrolled environment. The presented results are promising since most of the bunches were correctly detected with a 91% mean average precision, not only on the GrapeCS-ML database used to train the system, but also on an internal dataset, confirming the portability to different scenarios. Chapter 4 reports artichoke plant deep learning-based detection and georeferencing as the first step for an on-the-fly UAS spraying system and uses the gathered information to crop development monitoring in a multi-temporal approach. The Feature Pyramid Network, trained and compared with the YOLOv5 network, showed a high detection level with an average F1 score of around 90%, and satisfactory off-line performances on the Nvidia Jetson Nano board. The multi-temporal approach influenced detection performances, with an inverse response of precision and recall metrics. The growing index trend showed a distinct value in October, peaking at the beginning of December as expected

Machine learning and Unmanned Aerial Systems for crop monitoring and agrochemicals distribution optimization in orchard and horticultural systems / Sassu, Alberto. - (2023 Apr 14).

Machine learning and Unmanned Aerial Systems for crop monitoring and agrochemicals distribution optimization in orchard and horticultural systems

SASSU, Alberto
2023-04-14

Abstract

The work aims at discovering the potential and the efficiency of Unmanned Aerial Systems (UAS) and Machine Learning (ML) in agriculture scenario, focusing on crop management and agrochemicals distribution optimization in orchard and horticultural cropping systems. The dissertation includes a general introduction, three experimental chapters and a general conclusion. Chapter 2 illustrates an operational approach to estimate individual and aggregate vineyards’ canopy volume using the manual Tree-Row-Volume (TRV) and the remotely sensed Canopy Height Model (CHM) techniques, processed with MATLAB scripts, and validated through ArcGIS tools. The results confirm how the extensive use of TRV is recommended when supported by remote sensing, to better qualify errors and heterogeneities in field estimates. Chapter 3 presents the development of a grape bunch detector based on a deep convolutional neural network trained to work directly on the field in an uncontrolled environment. The presented results are promising since most of the bunches were correctly detected with a 91% mean average precision, not only on the GrapeCS-ML database used to train the system, but also on an internal dataset, confirming the portability to different scenarios. Chapter 4 reports artichoke plant deep learning-based detection and georeferencing as the first step for an on-the-fly UAS spraying system and uses the gathered information to crop development monitoring in a multi-temporal approach. The Feature Pyramid Network, trained and compared with the YOLOv5 network, showed a high detection level with an average F1 score of around 90%, and satisfactory off-line performances on the Nvidia Jetson Nano board. The multi-temporal approach influenced detection performances, with an inverse response of precision and recall metrics. The growing index trend showed a distinct value in October, peaking at the beginning of December as expected.
14-apr-2023
The work aims at discovering the potential and the efficiency of Unmanned Aerial Systems (UAS) and Machine Learning (ML) in agriculture scenario, focusing on crop management and agrochemicals distribution optimization in orchard and horticultural cropping systems. The dissertation includes a general introduction, three experimental chapters and a general conclusion. Chapter 2 illustrates an operational approach to estimate individual and aggregate vineyards’ canopy volume using the manual Tree-Row-Volume (TRV) and the remotely sensed Canopy Height Model (CHM) techniques, processed with MATLAB scripts, and validated through ArcGIS tools. The results confirm how the extensive use of TRV is recommended when supported by remote sensing, to better qualify errors and heterogeneities in field estimates. Chapter 3 presents the development of a grape bunch detector based on a deep convolutional neural network trained to work directly on the field in an uncontrolled environment. The presented results are promising since most of the bunches were correctly detected with a 91% mean average precision, not only on the GrapeCS-ML database used to train the system, but also on an internal dataset, confirming the portability to different scenarios. Chapter 4 reports artichoke plant deep learning-based detection and georeferencing as the first step for an on-the-fly UAS spraying system and uses the gathered information to crop development monitoring in a multi-temporal approach. The Feature Pyramid Network, trained and compared with the YOLOv5 network, showed a high detection level with an average F1 score of around 90%, and satisfactory off-line performances on the Nvidia Jetson Nano board. The multi-temporal approach influenced detection performances, with an inverse response of precision and recall metrics. The growing index trend showed a distinct value in October, peaking at the beginning of December as expected
CHM; grape detection; spraying UAS; temporal tracking; plant detection
Machine learning and Unmanned Aerial Systems for crop monitoring and agrochemicals distribution optimization in orchard and horticultural systems / Sassu, Alberto. - (2023 Apr 14).
File in questo prodotto:
File Dimensione Formato  
PhD_thesis_Alberto_Sassu.pdf

accesso aperto

Descrizione: Machine learning and Unmanned Aerial Systems for crop monitoring and agrochemicals distribution optimization in orchard and horticultural systems
Tipologia: Tesi di dottorato
Dimensione 4.43 MB
Formato Adobe PDF
4.43 MB Adobe PDF Visualizza/Apri

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11388/305707
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus ND
  • ???jsp.display-item.citation.isi??? ND
social impact