Skip to main content

Table 7 Comparison of different disease severity assessment methods in relation to accuracy-related measures, statistical methods and scale type and resolution used for method validation

From: From visual estimates to fully automated sensor-based measurements of plant disease severity: status and challenges for improving accuracy

MethodActual valueScale type and number of classes used for comparisonResolution to differentiate severityStatistic used to assess accuracyRangeReference
Visual assessmentImages traced from slide projections for contrast and image analyzedRatio scale
0 to 100% (100)
100Regression, intercept and slope coefficient (a and b)R2 = 0.913 to 0.960;
a = 1.064 to 12.958;
b = 1.029 to 1.245
Martin and Rybicki (1998
Images on paper cut by symptom border and weighedRatio scale
0 to 100% (100)
100LCC (ρc)ρc = 0.51–0.99Nita et al. (2003)
Manual image analysisRatio scale
0 to 100% (100)
100Regression, intercept and slope coefficient (a and b)R2 = 0.51 to 0.93;
a = −1.90 to 41.38;
b = 0.65 to 1.24
Godoy et al. (2006)
Manual image analysisRatio scale
0 to 100% (100)
100Regression, intercept and slope coefficient (a and b)
LCC (ρc)
Depending on symptom types:
R2 = 0.59 to 0.88;
a = − 1.52 to 2.83;
b = 0.10 to 1.21;
ρc = 0.85 to 0.94
Bock et al. (2008a)
Manual image analysisRatio scale
0 to 100% (100)
100Regression, intercept and slope coefficient (a and b)R2 = 0.88 to 0.98;
a = −6.68 to 5.09;
b = 0.75 to 0.94
Michereff et al. (2009)
Manual image analysisRatio scale
0 to 100% (100)
100LCC (ρc)Means of ρc = 0.76 to 0.98 depending on fruit perspective and use of SADsSpolti et al. (2011)
Manual image analysisRatio scale
0 to 100% (100)
100LCC (ρc)Mean of ρc = 0.79 and 0.89 (with and without SADs)Yadav et al. (2013)
Manual image analysisRatio scale
0 to 100% (100)
100LCC (ρc)ρc = 0.83 to 1.00 (mean = 0.95)Bardsley and Ngugi (2013)
Manual image analysisRatio scale
0 to 100% (100)
100LCC (ρc)Means of ρc = 0.53, 0.87, 0.86 and 0.87 (without SADs, with SADs, and with color or black and white SADs)Schwanck and Del Ponte (2014)
VIS (RGB) image analysisImages on paper cut by symptom border and weighedRatio scale
0 to 100% (100)
100Regression, intercept and slope coeffficient (a and b)R2 = 0.996;
a = −0.91;
b = 0.99
Lindow (1983)
Planimeter measurement, various pathosystemsRatio scale
0 to 100% (100)
100Regression, intercept and slopeR2 = 0.976 to 0.992;
a = 0.914 to 1.06;
b = − 0.17 to −4.35
Lindow and Webb (1983)
Images traced from slide projections for contrast and image analyzedRatio scale
0 to 100% (100)
100Regression, intercept and slopeR2 = 0.971 to 0.985;
a = − 0.877 to 0.610;
b = 0.999 to 1.045
Martin and Rybicki (1998)
Manual image analysisArea in pixelsRegression, intercept and slopeR2 = 0.980;
a = 0.901;
b = 16,097
Peressotti et al.( 2011)
Severity measured with multiplex real-time PCRRatio scale
0 to 100% (100)
100RegressionR2 = 0.9945De Coninck et al.( 2012)
Visual (pixels)Ratio scale
0 to 100% (100)
100Accuracy (%)Overall accuracy = 96%Barbedo (2014)
Visual ratings by 16 raters (inspection deemed image analysis was accurate)Ratio scale
0 to 100% (100)
100LCC (ρc)ρc = 0.76–0.99 (mean = 0.92)Stewart and McDonald (2014)
Visual (Pixels)Ratio scale
0 to 100% (100)
100PCC (r)r = 0.60–0.90Clément et al.( 2015)
Manual segmentation using PhotoshopRatio scale
0 to 100% (100)
100Quality of segmentation (Qs)Qs = 84.17%Hu et al. (2017)
VisualOrdinal (4 classes)Healthy stage, early stage, middle stage, end stageClassification accuracy (%)Classification accuracy:
Healthy stage = 100%;
Early stage = 93.1%;
Middle stage = 83.3%;
End stage = 97.0%
Wang et al.( 2017)
VisualOrdinal (5 classes)Healthy, very low, low, high, very highClassification accuracy (%) compared to other diseases and severitiesAccuracy of severity measurement = 78.57–86.51% (depending on architecture of CNN)Esgario et al. (2019)
VisualOrdinal (2 classes)Mild, severeClassification accuracy (%) compared to other diseases and severitiesSevere symptoms = 70.4%;
Mild symptoms = 29.4%
Ramcharan et al. (2019)
VisualOrdinal (3 classes)Healthy, general, seriousProportion accurately classified0.91Liang et al. (2019)
Multspectral (MSI) and Hyperspectral (HSI)VisualOrdinal (3 classes)Low, medium highClassification accuracy (%)71 to 91%, depending on class (a 5-class scale had accuracy = 11 to 40%)Coops et al. (2003)
VisualRatio scale
0 to 100% (100)
100Percentage results with error ≥ 5%24.1%Larsolle and Muhammed (2007)
VisualOrdinal (9 classes)0, 1, 10, 20, 30, 45, 60, 80% or 100%Regression, intercept and slopeR2 = 0.91;
a = 2.40;
b = − 721.22
Huang et al. (2007)
VisualOrdinal (4 classes)Severe, medium, light, non-visibleNone givenNone givenCui et al. (2009, 2010
VisualOrdinal (3 classes)Healthy tissue, light mycelium, dense myceliumClassification accuracy (%) including 3 diseases and their severitiesHealthy tissue = 100%;
Overall accuracy with disease = 61.70 to 98.90%
Mahlein et al. (2012b)
Symptom progressionChanging symptoms related to spectral changesMetro mapsWahabzada et al. (2015)
Symptom progressionChanging symptoms related to spectral changesLeaf traces (similar to above)Kuska et al. (2015)
VisualOrdinal (9 classes)0, 1, 5, 10, 20, 40, 60, 80 or 100Regression (R2)
RMSE
R2 > 0.90;
RMSE< 0.15
Wang et al. (2016)
Visual (in-field disease incidence of infected wheat spikes)Ratio scale
0 to 100% (100)
100Regression, intercept and slope (depending on VI)R2 = 0.801, 0.828;
a = 0.2902, 0.4572;
b = 0.0013, 0.0020
Kobayashi et al. (2016)
DNA quantification (presymptomatic)ContinuousDNA contentRegression (R2)R2 = 0.868Zhao et al. (2017)
DNA quantification (presymptomatic to symptomatic)ContinuousDNA contentRegression (R2)R2 = 0.72Thomas et al. (2017)
Length of lesionmmmmPredicted lesion length was proportional to the interior lesion length.Nagasubramanian et al. (2017)
VisualOrdinal (3 classes)Low (≤5%), moderate (5 to 20%), severe (> 20%) severity.Classification accuracy (%)94.83%Thomas et al. (2018a, b)
\