A Transforming Visual Analysis Approach for the Recognition of Colletotrichum Falcatum Leading to Red Rot of Sugarcane
https://doi.org/10.5281/zenodo.18051459
DOI:
https://doi.org/10.63075/pacxkk36Keywords:
AI, CNN, DNN, DRNN, KNN, RPN, Neural Network, Principal Component Analysis, and Support Vector MachineAbstract
The identification of Colletotrichum Falcatum in sugarcane has become a major global concern. A major contributor to sugarcane diseases is Colletotrichum Falcatum that triggers red rot disease in sugarcane. Multiple techniques are being developed to detect this type of disorder. Earlier studies have adopted k-nearest neighbors (KNN), Support Vector Machine (SVM), and Neural Network (NN) to discover Collectotrichum Falcatum red rot disease in sugarcane. The approaches currently in development for detecting Collectorichum falcutm are inefficient. These methods need the extraction of characteristics from images of sugarcane, which is used to identify Collectotrichum falciparum disease. Red rot infection was discovered in 88.2% of the measured fields. An estimate of the total losses in sugarcane output was 20.1%.[1]
This study suggests a fast and efficient image-based approach called Faster Region-based Convolutional Neural Network (R-CNN) for the identification of Collectotrichum Falcatum sugarcane infections. Region Proposal Network (RPN) fixes selective search, whereas Faster R-CNN handles it. Convolutional Neural Network (CNN) is used to first extract feature maps from the input image, which are then sent through an RPN to provide object recommendations. The datasets of sugarcane images are available at GitHub, which have approximately ninety thousand photographs available for free, are the source of the sugarcane datasets. Not many photos are taken directly using a cell phone. The approach is trained and tested using these images. A comparison is made between the proposed approach and K Nearest Neighbor (KNN). The experimental findings demonstrate that the suggested method outperforms the current method. Faster R-CNN If compared to KNN, sugarcane disease detection is more effective. Faster R-CNN achieved a detection rate of 0.96, whereas KNN produced 0.91. Furthermore, accuracy climbed to 0.98 whereas KNN generated 0.83.
