Table 12 Machine learning for soybean disease detection
From: Soybean cyst nematode detection and management: a review
Model | Objective | Dataset | Technology | Collection period | Spectrum range | Performance |
|---|---|---|---|---|---|---|
CNN [127] | Identifying SCN Egg Count | Soil Samples | Microscopic Imaging | Fall 2015 | N/A | ADA(94.33%), AMER(18.18%), AND(99.7%) |
GIS + RS [54] | Identifying SCN | Soybean Field near Ames, Iowa, in 2000 | Satellite Images and Aerial Images | 5 Collection Dates | 810 nm | Aerial Images (80%), Satellite Images (47%) |
SVDD [135] | Identifying Insect Damage | 100 Soybean Samples Harvested from a Garden in Zhejiang Province, China | Hyperspectral Imaging | 2011 Harvest Season | 400–1000 nm | 97.3% (normal), 87.5% (Insect-damaged) |
Ensemble ML [137] | Differentiating soybeans Seeds | 462 bands (25 varieties and 50 Seeds for Each) 1250 spectral curves | Hyperspectral imaging | N/A | 400–1000 nm | RSLD (99.2%), LD (98.6%), LSTM (69.7%) |
CNN [125] | Identifying SCN Egg count | Data collected from two fields in the State of Iowa | Microscopic images | Fall 2015 and Spring 2016 | N/A | Accuracy (95%), Average precision (93.73), F1-score (0.944) |
LDA, LgDA, and LCA [100] | Identifying SCN & SDS | Data collected (800 Leaf Spectra) from (Analytical spectral devices, Boulder, CO, USA) | Spectroscopic analysis in the NIR Region | Weekly basis, 71 days after planting (DAP) | 350–1070 nm | 97% (Healthy Plants) and 58% (Infested Plants)% |
Pixel-wise CNN [126] | Differentiating soybeans seeds | 3 Varieties of soybeans were prepared with 1890 soybeans in each variety | Hyperspectral imaging | 2019 | 975–1646 nm | Average accuracy (86%) |