Different Crop Leaf Disease Detection Using Convolutional Neural Network

Abstract

Crop diseases are a considerable danger to the crop's health, affecting the yield. Timely detection is challenging due to a lack of infrastructure in many regions of the world. Since they result in the death of plants, the loss of their product, and the global food problem, plant diseases must be investigated. Crop disease detection has been made possible by recent advancements in computer vision, deep learning, and the growing worldwide adoption of smartphones. Convolutional Neural Networks have significantly improved classifying images in the past several years. The performance of deep learning-based techniques for plant disease recognition under actual circumstances is thoroughly examined in this research. The objective was to offer some principles for conducting a more thorough and realistic examination of deep learning-based approaches for disease recognition. Sequential Architecture was used to classify 38 diseases of 14 crops on a crop leaves image dataset containing 70,295 training and 17,572 testing images. A simple convolutional neural network has been proposed that detects crop diseases seamlessly. The maximum accuracy obtained was 95% on the 14 th epoch. This was accomplished by following the Sequential Model. It is a cuttingedge network that can help new researchers who desire to conduct their studies in deep learning applications with an emphasis on agriculture.

Enzyme dosage detection to degrade feathers in edible bird’s nests: A comparative convolutional neural networks study

“…This is due to the role of validation accuracy, which evaluates the model's performance in classifying the unseen data during the training process, with the formula, as shown in Equation 4. Therefore, among the four-performance metrics, the validation accuracy was prioritized for selecting the best model since the ultimate goal of machine learning is to produce a model that can perform well and accurately classify unseen data, making it more robust and relevant to real-world scenarios (Pawar et al, 2023). Validation loss is a metric used to evaluate the model's performance in minimizing errors or loss on the validation data.…”