Novel Pyramidal Bidirectional Gated Vision Transformer for Rice Leaf Disease Detection
Worldwide, rice is a vital crop, but it is often affected by disease during the growth process. Rice leaf diseases include flax leaf spot, rice blast, and bacterial blight. These illnesses were very infectious and lethal, and they might pose a significant barrier to agricultural progress. The rice l...
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Ital Publication
2025-06-01
|
| Series: | Journal of Human, Earth, and Future |
| Subjects: | |
| Online Access: | https://hefjournal.org/index.php/HEF/article/view/493 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Worldwide, rice is a vital crop, but it is often affected by disease during the growth process. Rice leaf diseases include flax leaf spot, rice blast, and bacterial blight. These illnesses were very infectious and lethal, and they might pose a significant barrier to agricultural progress. The rice leaf image contains noise and unclear edges, which can affect the accurate extraction of disease. However, because rice leaf diseases might be identical to one another, classifying photos of diseases can be a very challenging task. The categorization of rice leaf diseases using a hybrid deep learning (DL) approach with an efficient feature extraction technique is innovative in this research. The three stages of this paper are feature extraction, classification, and pre-processing. First, the image from the dataset is extracted, and the adaptive Gaussian bilateral filter (AdGaBF) is used to remove noise. The features are then extracted, and the disease is classified using the deep multi-scale feature enhanced pyramidal convolutional block assisted bidirectional gated transformer (Dep-MPc-BgT) technique. The experiment uses two datasets, the rice leaf disease image dataset and the rice leaf disease dataset, and the approach successfully diagnoses the disease. Both datasets effectively evaluate the performance by using various metrics such as accuracy, precision, recall, F1-score and kappa score, MSE, RMSE, and processing time. The proposed rice leaf image dataset obtained values of 99.23%, 99.02%, 98.89%, 98.67%, 98.57%, 0.964, 0.991, and 0.118 s correspondingly. Similarly, the rice leaf dataset also achieved better performance of accuracy, precision, recall, F1-score, Kappa score, MSE, RMSE, and processing time values of 99.56%, 98.23%, 99.34%, 99.16%, 98.67%, 0.972, 0.994, and 0.127 s, respectively. |
|---|---|
| ISSN: | 2785-2997 |