Abstract
Chile is one of the main exporters of sweet cherries in the world and one of the few in the southern hemisphere, being their harvesting between October and January. Hence, Chilean cherries have gained market in the last few years and positioned Chile in a strategic situation which motivates to undergo through a deep innovation process in the field. Currently, cherry crop estimates have an error of approximately 45%, which propagates to all stages of the production process. In order to mitigate such error, we develop, test and evaluate a deep neural-based approach, using a portable artificial vision system to enhance the cherries harvesting estimates. Our system was tested in a cherry grove, under real field conditions. It was able to detect cherries with up to 85% of accuracy and to estimate production with up to 25% of error. In addition, it was able to classify cherries into four sizes, for a better characterization of the production for exportation.
Highlights
Chile produced more than 100,000 tonnes of sweet cherries in 2017, which makes it the largest producer of cherries in the southern hemisphere [1]
To analyse the usability of Faster R-CNN for cherry harvesting estimation, 20% of the imagery database was used for validation purposes and the remaining 20% for testing
70% of the images presented this underestimation, this data could be considered in future work to make a compensation in the estimation of production
Summary
Chile produced more than 100,000 tonnes of sweet cherries in 2017, which makes it the largest producer of cherries in the southern hemisphere [1]. Such an amount of production requires the optimization of processes related to the stages of sowing, production monitoring and harvesting, the need for technological advances in the field. [5] studied the feasibility of using computer vision to rate cherry color under outdoor and natural light conditions, where the maturity of cherries can be classified according to NIR-hyperspectral imaging [6]. Hyperspectral cameras are used in cherry crops, as it is shown in [6], where a NIR (near-infrared) hyperspectral imagery system was used to classify the maturity of cherries from their spectral signature
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
