Skip to main content

Table 1 Comparison of classification accuracies of DeepSort and other classifiers. Other classifiers were tested with all features described in text (values within brackets), and using only Haralick texture features (values outside brackets). CV indicates 5-fold cross-validation

From: DeepSort: deep convolutional networks for sorting haploid maize seeds

  DeepSort Random Forest SVM Logistic Reg
CV 0.961 0.840 (0.823) 0.857 (0.836) 0.749 (0.777)
Train 1.000 1.000 (0.997) 0.911 (0.994) 0.751 (0.786)
Test 0.968 0.845 (0.824) 0.876 (0.839) 0.775 (0.772)
\