Grape Leaf Diseases Identification System Using Convolutional Neural Networks and LoRa Technology

Image transmission over Low-Power Wide Area Networks (LP-WAN) protocols has always been a difficult task since it necessitates high data rates and energy consumption. Long Range (LoRa) is one such protocol, which excellent for transferring long distances but generated severe doubts regarding the viability of image due to its low rate. This paper demonstrates application results an integrated LoRa Deep Learning-based computer vision system that can efficiently identify grape leaf diseases using low-resolution images. In particular, focus in this combine two technologies, Learning, make images identification possible. To achieve objective, framework utilizes combination on-site simulation experiments along with different parameters Convolutional Neural Model (CNN) model fine-tuning. Based on evaluation, proposed proved possible within protocol limitations (such as limited bandwidth duty cycle). Our fine-tuned leaves diseases. The technique both efficient adaptive specifics each disease, while does not need any training adjust parameters. It worth noting today, end-user trust Machine Learning models increased significantly because novel solutions field Explainable Artificial Intelligence (XAI). study, we use Grad-CAM method visualize output layer judgments CNN. disease’s spot region highly activated, according visualization findings. how network distinguishes between