Performance of the SVM phenotype recognition model in 10-fold cross-validationa.

aThe overall recognition accuracy is 97.40±0.95%.bThe class precision is the percentage of correct classified samples in a predicted class. For example the precision of the (predicted) hatched class is given by 524/(524+4+8) = 97.76%.cThe class recall is the proportion of the samples in the class that were correctly identified. For example the recall of (true) hatched class is 524/(524+4+0) = 99.24%.a. 整体识别准确率为97.40±0.95%。 b. 类别精确率指某一预测类别内被正确分类的样本占比。示例：（预测为）阴影类别的精确率为524/(524+4+8)=97.76%。 c. 类别召回率指某一真实类别内被正确识别的样本占比。示例：（真实为）阴影类别的召回率为524/(524+4+0)=99.24%。