Precision Disease Grading of Botrytis Cinerea in Kiwifruit Leaves Via CNNs and Logistic Regression

The abstract gives a short description of the test that was done on how well a classification model worked at ten different gravity levels. For each level, precision, recall, or Fl-Score were calculated, which showed that forecasts were always very accurate. It got Fl scores between 93.37% and 96.65%, with levels 4, 6, and 7 of gravity showing especially good results, all of which were above 96%. The precision values were always higher than 95%, which meant that fake positives were very rare, and the recall values showed how well the model worked at finding true positives. Support values, which show the number of samples in each class, showed that there was a lot of data across all gravity levels, which meant that the forecasts could be trusted. The collection of 16,375 pictures was used to test a classification model through ten gravitational levels. The model achieved an average accuracy of 90.07%. The model also showed an impressive total accuracy of 99%, which makes it even more reliable for correctly classifying gravitational levels according to the information that was given. All the results indicate that the model functions properly, and might be employed in cases where strict sorting of gravitational levels is required. In conclusion of the overview of the model, it is pointed out that the model is suitable for those tasks which require precise classification and that the model has been subjected to various rigorous tests and has performed well on numerous sundry complex classification tasks.