An Optimized Resnet based Plant Disease Identification using Deep Learning Hypothetic Function

Since the world is pragmatically changing with the deep learning approaches one such research is trending in the field of apple plant disease identification, whose adoption in developing nations may be challenging as a very few investigations done on apple cultivation which is very crucial fruit which to be explored in this research. Starting with the images captured with the sufficient resolutions is typically used for training existing deep learning models. However, models that would work effectively even with low-resolution data are required for developing nations with sparse internet access. The created models are quite large in number which results the majority of deep learning-based apps are accessible online. Unfortunately, such larger models required large memory requirements which are becoming a new problem. Therefore, smaller-sized models being recommended. In this paper, plant disease diagnosis using low-resolution data is accomplished using an existing deep learning model called Resnet. Deep convolution neural networks (DCNN), which are frequently used for image data, are the subject of the main attention. Concatenated residual networks, skip connections, and a novel DCNN architecture are utilized for the accurate results. by vanishing gradient with training parameters and making the with the hypothetical function. The study focused to treat the apple disease identification whose data collected from kaggle wherein the images consists of 10 classes and 5 classes of diseases that are frequently seen in Dhaka, Bangladesh, as well as a healthy class, are also used to assess the performance of the Plant Village dataset. Our proposed resnet 38 with the tuning hypothetical function justifies the significance of the overall approach whith the accuracy of 99.62% that is better than the cited DCNN networks.