Empowering Agricultural Insights: RiceLeafBD - A Novel Dataset and Optimal Model Selection for Rice Leaf Disease Diagnosis through Transfer Learning Technique

Sadia Afrin Rimi, Md. Jalal Uddin Chowdhury, Rifat Abdullah, Iftekhar Ahmed, Mahrima Akter Mim, Mohammad Shoaib Rahman · 2025

Read Empowering Agricultural Insights: RiceLeafBD - A Novel Dataset and Optimal Model Selection for Rice Leaf Disease Diagnosis through Transfer Learning Technique by Sadia Afrin Rimi, Md. Jalal Uddin Chowdhury, Rifat Abdullah, Iftekhar Ahmed, Mahrima Akter Mim, Mohammad Shoaib Rahman (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2501-08912.txt.

Introduction

Six researchers from Bangladesh introduce RiceLeafBD, a new image dataset for spotting diseases in rice leaves, and test which computer model reads it best. Rice feeds more than half the world, yet common leaf diseases cut yields year after year, and early detection is the hardest part of control. The team collected its own field images in Bangladesh and trained several deep learning models on them.

A light custom network and three pretrained models were compared through transfer learning, in which a model trained on general images is retuned for a new task. The EfficientNet-V2 model reached 91.5 percent accuracy, ahead of the others and of earlier published approaches. The authors argue that an unbiased, locally gathered dataset plus a well-chosen model can identify rice leaf diseases precisely, and they offer the data as a base for further work.

Catalog record

CREATORSadia Afrin Rimi, Md. Jalal Uddin Chowdhury, Rifat Abdullah, Iftekhar Ahmed, Mahrima Akter Mim, Mohammad Shoaib Rahman
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE38 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2501-08912

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Empowering Agricultural Insights: RiceLeafBD - A Novel Dataset and Optimal Model Selection for Rice Leaf Disease Diagnosis through Transfer Learning Technique Sadia Afrin Rimi, Md. Jalal Uddin Chowdhury, Rifat Abdullah, Iftekhar Ahmed, Mahrima Akter Mim, Mohammad Shoaib Rahman Abstract The number of people living in this agricultural nation of ours, which is surrounded by lush greenery, is growing on a daily basis. As a result of this, the level of arable land is decreasing, as well as residential houses and industrial factories. The food crisis is becoming the main threat for us in the upcoming days. Because on the one hand, the population is increasing, and on the other hand, the amount of food crop production is decreasing due to the attack of diseases. Rice is one of the most significant cultivated crops since it provides food for more than half of the world’s population. Bangladesh is dependent on rice (Oryza sativa) as a vital crop for its agriculture, but it faces a significant problem as a result of the ongoing decline in rice yield brought on by common diseases. Early disease detection is the main difficulty in rice crop cultivation. In this paper, we proposed our own dataset, which was collected from the Bangladesh field, and also applied Deep learning and Transfer learning models for the evaluation of the datasets. We elaborately explain our dataset and also give direction for further research work to serve society using this dataset. We applied a light CNN model and pre-trained InceptionNet-V2, EfficientNet-V2, and MobileNet-V2 models, which achieved 91.5% performance for the EfficientNet-V2 model of this work. The results obtained assaulted other models and even exceeded approaches that are considered to be part of the state of the art. It has been demonstrated by this study that it is possible to precisely and effectively identify diseases that affect rice leaves using this unbiased datasets. After analysis of the performance of different models, the proposed datasets are significant for the society for research work to provide solutions for decreasing rice leaf disease. keywords Rice Leaf Disease, Deep Learning, Transfer Learning, Proposed Datasets, and RiceLeafBD † † equal-contributors: These authors contributed equally to this work. † † equal-contributors: These authors contributed equally to this work. † † equal-contributors: These authors contributed equally to this work. † † equal-contributors: These authors contributed equally to this work. † † equal-contributors: These authors contributed equally to this work. † † equal-contributors: These authors contributed equally to this work. 1 Introduction As the backbone of economic stability, agriculture is of utmost importance, especially in countries such as Bangladesh, where it is the primary contributor to GDP. An essential food crop, rice farming is critical to maintaining economic stability and creating significant job opportunities Islam et al. 2022 . Nonetheless, the agricultural industry has significant obstacles, chief among which is the enduring danger of illnesses, which significantly jeopardizes rice farming and influences the amount and quality of output Yusrin 2023 . Traditionally, disease diagnosis techniques that rely on subjective visual cues or time-consuming laboratory testing highlight subjectivity, deadlines, and a reliance on expert knowledge Panayides et al. 2020 . In light of these constraints, modern agricultural research is gradually incorporating cutting-edge technology with an emphasis on the revolutionary potential of computer vision and machine learning to improve and automate disease identification procedures Shaikh et al. 2022 . In the last few decades, a lot of work has gone into collecting data. Several sets of data have been acknowledged. PlantVillage Hughes et al. 2015 , which has 54,000 images of the underside side of leaves on a plain background, is the most well-known in this area. But, as has been seen in other studies Barbedo 2019 , these combinations fail to demonstrate that the goals of the end program are adequate. The datasets that were made under controlled conditions—that is, with the leaf on a background that is all the same—do not accurately reflect the real-world conditions in which the model will work. The importance of this study has been created clear on two levels in this circumstance. We add a new dataset to the literature called RiceLeafBD that can be used to diagnose and keep an eye on plant constraints. It consists of a set of 1555 images taken in real-life fields that show four types of leaf stress: Leaf Blight, Tungro virus, Brown Spot, and healthy leaves. We are doing a study of public images that were made to help classify and identify leaf diseases. We focus on datasets that are available on sites for sharing data in an open way because now, a time-challenging thing is open access qualified datasets. So, we don’t deal with files that are given to writers upon request. After publishing this study, we will make datasets available to the public, which is helpful in two ways. First, it saves researchers time and money, which lets them put more effort into comparing and evaluating algorithms in a manner that is less biased. We think that this gathering will help researchers choose the best samples for subsequent research. This study also investigates the detection of rice leaf disease used own dataset for the context of Bangladesh’s distinctive agricultural terrain. The application of cutting-edge deep learning models, such as light CNN, MobileNet-v2, and InceptionNet-v2, represents a tactical shift toward the use of artificial intelligence for accurate and timely disease diagnosis and also proves these model’s performance that datasets are not biased. This study is novel because it directly affects food security and economic results, both of which are significantly impacted by early illness diagnosis with the use of the actual region’s own datasets. The structure of this document is as follows: Section 2 provides an overview of the Literature Review used in this research study. The Datasets Description is proposed in Section 3. Section 4 covers the materials and techniques, including the algorithms and datasets utilized. Section 5 provides a summary of the results obtained from this study. The Conclusion and future work are presented in Section 6. 2 Literature Review The study of some recent image-processing research works is presented in this section. This review of the literature discusses a number of approaches and strategies for using image processing to detect rice leaf disease. This study considers many algorithmic models to account for the implications of machine learning and deep learning approaches. A few recent research on the identification and classification of diseases affecting rice plants are included in the list below. Pallapothula Tejaswini et al. Tejaswini et al. 2022 classified diseases affecting rice leaves. Their suggested 5-layer convolutional network outperforms the other common deep learning models in accuracy by about 6%. They discovered that they could get a notable level of accuracy with a handcrafted model with fewer layers by modifying the training parameters, such as learning rate, epochs, and optimizer approaches. As they get more adept at identifying Daffodil International University The easier it will be for farmers to safeguard their crops against these six illnesses. With 78.2% accuracy, the 5-layer convolution model performed the best.Based on the disease stage, some ideas for fertilizers and pesticides are made in this study, which uses an SVM classifier to identify rice ailments. In terms of

Read the complete text