Answers are grounded in this paper and cite their sources.
A transfer learning-based, 5G-enabled, under-canopy sugarcane protection with insect and disease classification network (SPIDCNet) is proposed. Online sugarcane datasets pose several challenges: incorrect labels, limited class labels, and few images per class. To overcome the limitations of online datasets, a custom sugarcane dataset is curated, comprising 4500 images across five classes: Healthy, Grassy shoot, Yellow leaf disease (YLD), Ring spot, and Web mites. Experts in sugarcane diseases and insects validated our custom dataset. The SPIDCNet is built over a set of fine-tuned deep-learning models and a novel insect and diseases classification head. The deep-learning models MobileNetV2, InceptionV3, ResNet50, VGG16, EfficientNetV2B1, and Xception, fine-tuned on our custom sugarcane dataset, achieved testing accuracies of 97.64%, 98.71%, 98.93%, 94.23%, 98.71%, and 96.79%, respectively. An interactive 5G-enabled farmer mobile application was developed to connect sugarcane farmers to SPIDCNet hosted on our web server through the public network. The farmer mobile application performs well on 5 G even on a 4 G network, in terms of network latency. The proposed system enables farmers to upload sugarcane images from their fields and get expert guidance to manage sugarcane insects and diseases to improve crop yield.
No takes yet. Share an insight, caveat, or question.
S et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: