Click on “Download PDF” for the PDF version or on the title for the HTML version.

If you are not an ASABE member or if your employer has not arranged for access to the full-text, Click here for options.


Published by the American Society of Agricultural and Biological Engineers, St. Joseph, Michigan

Citation:  Paper number  021169,  2002 ASAE Annual Meeting . (doi: 10.13031/2013.10586) @2002
Authors:   Swapna Gogineni, Jeffrey G. White, J. Alex Thomasson, Paul G. Thompson, James R. Wooten, Mark Shankle
Keywords:   Sweetpotatoes, Precision agriculture, yield monitoring, Image analysis, Neural networks, Grading

An image-based system for monitoring yield and grade of sweetpotatoes was developed. Estimates of weight were based on multiple-linear regression and neural networks, while grade classifications were based on linear discriminant analysis and neural networks. Sweetpotato features considered were pixel area, polar moment of inertia, rectangular height and width, and length of major and minor axes. The system was tested on stationary sweetpotatoes in the laboratory. Its estimates of sweetpotato weights were highly correlated (R2 = 0.96) with actual weights, and grade classifications of marketable sweetpotatoes were over 90% accurate. The system was also tested on sweetpotatoes moving on a harvesters conveyor belt in the field. In this portion of the study, estimates of sweetpotato weights were still highly correlated (R2 = 0.91), albeit not as strongly, with actual weights. Grade classifications during harvesting were less accurate (R2 = 0.73 in the best case) than in the laboratory.

(Download PDF)    (Export to EndNotes)