Horticulture Image Based Segmentation with Feature Selection Using U-ConVolNet with Boltzmann Machine Using Deep Learning Architectures

Abstract

The most significant role on Earth is played by plants. In both the ecological and medical fields, every organ of a plant is essential. However, there are many different plant species on the planet. Different diseases affect various plants. In order to avoid loss, it is necessary to identify the plants and their illnesses. Currently, it takes a lot of time to manually detect the diseases that affect plants. This study suggests a novel method for segmenting horticulture photos using feature selection using deep learning techniques. Here, the input image has undergone noise removal, smoothing, and normalisation processes after being gathered as horticultural images. Using U-ConVolNet and a Boltzmann machine, the processed picture has been segmented and features have been chosen. The experimental analysis has been done in terms of RMSE, MAP, F-1 Score, recall, accuracy, and precision. The proposal had 95% accuracy, 84% recall, 73% F-1 score, 53% RMSE, and 58% MAPE.