Open Access Open Access  Restricted Access Subscription or Fee Access

Rice Plant Disease Classification: Training using Transfer Learning, Fine Tuning or Scratch Learning?

Vimal K. Shrivastava, Monoj K. Pradhan, Mahesh P. Thakur

Abstract



Agriculture products and its productivity considerably contribute to the nation’s economy. Rice among other crops is a staple food and grown globally. However, various diseases affect the rice plants and cause a huge loss in both productivity and quality. Hence, rice plant disease detection in their early stage and their classification is an essential step to control its spread over the field by applying suitable techniques for sustainable agriculture. Currently, farmers are dependent on experts for this task which is time taking and prone to human error. Identification of diseases from images of rice plant is one of the important research area and machine learning concept can be applied for accurate, effective and fast detection. Recently, deep learning models specifically convolution neural network (CNN) has shown tremendous success result in image classification task. Motivated by this, we have explored 16 off-the-shelf deep CNN models and demonstrated their performances on classification of image-based diseases from rice plant. Further, the performance of these 16 models were compared with three approaches: transfer learning, fine tuning and scratch learning. Here, we have considered 1216 images collected from the real agriculture field that belongs to seven classes: (a) Rice Blast, (b) Bacterial Leaf Blight, (c) Sheath Blight, (d) Brown Spot, (e) Sheath Rot, (f) False Smut and (d) Healthy Leave. DenseNet121 model obtained superior performance with an average classification accuracy of 98.36%. Hence, the analysis presented in this paper exemplifies that DenseNet121 model can be used as an advisory for early detection and classification of image based diseases from rice plant.

Keywords


Sustainable Agriculture, Rice Plant Disease, Artificial Intelligence, Convolution Neural Network, Off-the-Shelf Models, Classification.

Full Text:

PDF


Disclaimer/Regarding indexing issue:

We have provided the online access of all issues and papers to the indexing agencies (as given on journal web site). It’s depend on indexing agencies when, how and what manner they can index or not. Hence, we like to inform that on the basis of earlier indexing, we can’t predict the today or future indexing policy of third party (i.e. indexing agencies) as they have right to discontinue any journal at any time without prior information to the journal. So, please neither sends any question nor expects any answer from us on the behalf of third party i.e. indexing agencies.Hence, we will not issue any certificate or letter for indexing issue. Our role is just to provide the online access to them. So we do properly this and one can visit indexing agencies website to get the authentic information. Also: DOI is paid service which provided by a third party. We never mentioned that we go for this for our any journal. However, journal have no objection if author go directly for this paid DOI service.