Sugarcane Severity Assessment with Deep Learning and Random Forest: A Multi-Class Approach

Researchers evaluate a classification method based on a thorough table of performance metrics in this abstract. Data is categorized using this method into five significant levels (Significance Levels 1 to 5), which stand for various classes. The model's precision regularly oscillates between 96.77% to 97.06%, demonstrating its accuracy along with a low false positive rate. Recall ratings, which range from 96.34% to 97.38%, further demonstrate the model's accuracy in identifying occurrences of each class. High recall with accuracy values is a byproduct of exceptional F1 scores, ranging from 96.59% to 97.13%, achieving a balance between precision and recall. The total amount of instances inside each class is shown by the support values, having class 5 receiving the most support (1025), indicating a possible class imbalance. The support proportion shows that a sizable chunk of the dataset belongs to class 5. Its overall performance is summarized by macro, micro, and overall weighted averages, all of which have a constant precision, recall, or F1-score of 96.88%. It achieves an outstanding overall accuracy of over 96.87% with a data set of 4,800 photos. A micro average accuracy, recall, & F1-score of 96.88% show how well the model categorizes cases across the dataset. Class differences are taken into consideration, and the weighted average accuracy stays at 96.88%. In conclusion, the categorization model routinely outperforms industry standards, demonstrating its ability to precisely categorize data into different categories and efficiently achieve its intended purpose.