Author
![]() |
ATOUM, YOUSEF - Michigan State University |
![]() |
AFRIDI, MUHAMMED JAMAL - Michigan State University |
![]() |
LIU, XIAOMING - Michigan State University |
![]() |
McGrath, Jon |
![]() |
Hanson, Linda |
Submitted to: Pattern Recognition
Publication Type: Peer Reviewed Journal Publication Acceptance Date: 11/28/2015 Publication Date: 3/9/2016 Citation: Atoum, Y., Afridi, M., Liu, X., McGrath, J.M., Hanson, L.E. 2016. On developing and enhancing plant-level disease rating systems in real fields. Pattern Recognition. 53:287-299. Interpretive Summary: Cercospora leaf spot is one of the most serious diseases in sugar beet plants, and can cause an epidemic leading to complete defoliation of the plant and loss of income to growers. Fungicides are effective, however in many cases, Cercospora has developed fungicide resistance, thus genetic resistance is highly desirable. Methods have been developed to assess the degree of disease exhibited by different varieties, however these are laborious and require extensive familiarity with the disease for accurate ratings of the degree of resistance. Thus, a method was desired that would use information from video capture of plants during the disease progression that would give both a numerical score and be reproducible across seasons, as well as could be accomplished in less time than current methods. Video images were analyzed on the basis of color and texture using different numbers of pixels (superpixels) and these superpixels were used to describe the amount of disease exhibited by a variety. This new method showed high correlation with the existing scoring system with good consistency, taking a quarter of the time required manually. Technical Abstract: Cercospora leaf spot (CLS) is one of the most serious diseases in sugar beet plants causing an enormous decrease in the sugar production throughout the world. Agricultural researchers are continuously seeking CLS-resistant sugar beet cultivars. Normally human experts manually observe and rate the resistance of a large variety of sugar beet plants over a period of a few months. Unfortunately, this procedure is laborious and subjective from one expert to another resulting in large disagreements on the level of resistance. Therefore, we propose a novel computer vision system, CLS Rater, to automatically and accurately rate plant images in the real field to the “USDA scale” of 0 to 10. Given a set of plant images captured by a tractor-mounted camera, CLS Rater extracts multi-scale superpixels, where in each scale a novel histogram of importances feature encodes both the within-superpixel local and across-superpixel global appearance variations. These features at different superpixel scales are then fused for learning a bagging M5P regressor that estimates the rating for each plant image. We further address the issue of the noisy labels by experts in the field, and propose a method to enhance the performance of the CLS Rater by automatically modifying the experts ratings to ensure consistency. We test our system on the field data collected from two years over a two-month period for each year, under different lighting and weather conditions. Experimental results show that both the CLS Rater and the enhanced CLS Rater to be highly consistent with the rating errors of 0.65 and 0.59 respectively, which demonstrates a higher consistency than the rating standard deviation of 1.31 by the human experts. |