Abstract Title
Video Processing for Fruit Yield Prediction
Additional Funding Sources
The project described was supported by a Specialty Crop Block Grant by the Idaho State Department of Agriculture.
Abstract
Blossom counting in fruit orchards has been identified as having the potential for estimating final crop yield. This is invaluable information for fruit farmers as they can market their crop earlier as well as prepare for the harvest more effectively. However, capturing and analyzing photos of each tree in the orchard is a time-consuming process. This study uses live video acquisition system attached to an autonomous vehicle to count blossoms as it goes. Both a simple color filtration segmentation and shallow neural network segmentation were used for detection. The results show potential for fruit yield estimation however there is a need to improve the blossom detection for variable lighting condition in the orchard.
Video Processing for Fruit Yield Prediction
Blossom counting in fruit orchards has been identified as having the potential for estimating final crop yield. This is invaluable information for fruit farmers as they can market their crop earlier as well as prepare for the harvest more effectively. However, capturing and analyzing photos of each tree in the orchard is a time-consuming process. This study uses live video acquisition system attached to an autonomous vehicle to count blossoms as it goes. Both a simple color filtration segmentation and shallow neural network segmentation were used for detection. The results show potential for fruit yield estimation however there is a need to improve the blossom detection for variable lighting condition in the orchard.
Comments
W40