Combining Convolutional Neural Networks and Random Forest for Lotus Multi-Classification

This study examines that, after doing an in-depth investigation the researchers found that the classification of lotus plants helps us to understand the taxonomic organization, plant breeding, and medicinal value. The current paper has used the hybrid approach that combines a convolutional neural network and a random forest. This data is collected from the area of Punjab. The paper describes all five classes with a minimum accuracy of 0.94% and a maximum accuracy of 0.98%. The total macro average for precision, recall and F1score is 90.84%, 90.50%, and 90.24%. The total micro average for the precision, recall, and F1-score is 90.33% respectively. Similarly, the total weighted average for precision is 91.51%, for recall it is 90.33% and for F1-score it is 90.53%. It is very helpful for farmers because it saves their time and costs. The lotus plants help farmers to prepare pickles or use them raw for salad-making by cutting and slicing. In the future, it will be helpful to find other hybrid approaches for better accuracy.