Building a Vineyard Information System for Individual Vines Using a Low-Cost Camera Xu Wang a , Paul Petrie a,b , Marcos Bonada b , Mark Whitty a a School of Mechanical and Manufacturing, UNSW Australia, NSW 2052, Australia b The South Australian Research and Development Institute, Waite Campus, Australia Introduction Gathering accurate information on individual vines can help growers manage intra-vineyard variability. Mounting a low cost camera with an integrated GPS receiver on vineyard vehicles allows an information system to be automatically maintained using off the shelf hardware as part of normal operations. The video is analysis to determine the number and location of vines within the block. Vine size (cordon length) and the proportion non-bearing canopy can be calculated and used as an input to yield estimation systems. The vine-level metadata is automatically generated for each vineyard pass and presented as a map of the block. The images and maps allow quick and easy monitoring of trunk diseases such as Eutypa and any other visual elements (e.g. broken posts) allow specific sites to be targeted. Methodology Results & Conclusion • In the example row 72 out of 73 vines and all of the posts were successfully detected. • A vine-level information system has been proposed in the form of a map which contains vine counts, location, and cordon lengths. • The location of missing vines can be highlighted from the map generated by the system. • This system can be later extended to collect other factors such as shoot counts and yield estimates. Two locations are detected for missing vines as there are only two trunks between posts and also very large gap between the cordons. See the image for the first position. Vine Location Vine GPS Vine Length Shoot Counts Row 2 Vine 18 Lon 34.4764 Lat -139.0170 1.98 m …. Vine Information Table The vineyard information system map shows both vines (blue dots) and posts (red dots). For each vine in a row, it contains a vine information table, showing vine number, location, vine length and other visually detected vine properties. In the figure above, there should be a post located every three vines, and therefore red squares denote missing vines. Start End GPS trace from low-cost camera Detected Post Detected Vine Cordon Detected Vine Trunk

Building a Vineyard In formation System for Individual Vine s … · 2019. 7. 9. · Building a Vineyard In formation System for Individual Vine s Using a Low-Cost Camera Xu Wanga,

Download PDF Report

Upload
others
View
0
Download
0

Embed Size (px)

Citation preview

Page 1: Building a Vineyard In formation System for Individual Vine s … · 2019. 7. 9. · Building a Vineyard In formation System for Individual Vine s Using a Low-Cost Camera Xu Wanga,

Building a Vineyard Information System for Individual VinesUsing a Low-Cost Camera

Xu Wanga, Paul Petriea,b, Marcos Bonadab, Mark Whittya

a School of Mechanical and Manufacturing, UNSW Australia, NSW 2052, Australiab The South Australian Research and Development Institute, Waite Campus, Australia

IntroductionGathering accurate information on individual vines can help growers manage intra-vineyardvariability. Mounting a low cost camera with an integrated GPS receiver on vineyard vehiclesallows an information system to be automatically maintained using off the shelf hardware aspart of normal operations. The video is analysis to determine the number and location of vineswithin the block. Vine size (cordon length) and the proportion non-bearing canopy can becalculated and used as an input to yield estimation systems. The vine-level metadata isautomatically generated for each vineyard pass and presented as a map of the block.The images and maps allow quick and easy monitoring of trunk diseases such as Eutypa andany other visual elements (e.g. broken posts) allow specific sites to be targeted.

Methodology

Results & Conclusion• In the example row 72 out of 73 vines and all of the posts were successfully detected.• A vine-level information system has been proposed in the form of a map which contains vine counts, location, and

cordon lengths.• The location of missing vines can be highlighted from the map generated by the system.• This system can be later extended to collect other factors such as shoot counts and yield estimates.

Two locations are detected formissing vines as there are onlytwo trunks between posts andalso very large gap between thecordons. See the image for thefirst position.

Vine Location

Vine GPS

Vine Length

Shoot Counts

Row 2 Vine 18

Lon 34.4764Lat -139.0170

1.98 m ….

Vine Information Table The vineyard information system map shows both vines (blue dots)and posts (red dots). For each vine in a row, it contains a vineinformation table, showing vine number, location, vine length andother visually detected vine properties. In the figure above, thereshould be a post located every three vines, and therefore redsquares denote missing vines.

Start

End

GPS trace from low-cost camera

Detected Post

Detected Vine CordonDetected Vine Trunk