Classification-of-Potato-Disease-A-Hybrid-Deep-Learning-Framework

Abstract

Potatoes are among the major vegetables in agricul- tural regions, and it is farmed and utilized all over the world. Potatoes are a high-protein food with several health benefits, but there are numerous diseases associated with potatoes that hamper production. In this research, we developed a hybrid approach that employs image processing and combines MobileNet V2 with LSTM, GRU, and Bidirectional LSTM to evaluate potato disease classes known as Black Scurf, Common Scab, Blackleg, Dry Rot, Pink Rot, Healthy, and Miscellaneous. We examined the outcomes of each architecture after applying it independently to determine the optimal architecture configuration for categorizing potato diseases. In terms of accuracy, the results show that the hybrid MobileNet V2-GRU with Stochastic Gradient Descent optimizer strategy exceeds the other alternative. On the test dataset, we achieved 99% accuracy.

Table of Contents

• Paper
• Proposed Methodology
• Dataset Availability
• Results
• Contact Information
• Citation

Paper Link

Explore our research on "Classification of Potato Diseases" to delve into our innovative hybrid deep learning framework for accurate disease classification. To view our detailed findings and methodologies, access the full paper here.

Proposed Methodology

Dataset Availability

The dataset can be accessed from here.

Specifics of the Core Data:

Disease Name	Total Images	Causes of Disease
Common scab	47	Bacteria
Blackleg	40	Bacteria
Dry rot	44	Fungus
Pink rot	46	Fungus
Black scurf	36	Fungus

Results

Table 1: Potential Assessment Predictions of Deep Hybrid Learning Model to Classify the Potato Disease using Adam Optimizer

Decoder	Accuracy	Precision	Recall	F1 Score
LSTM	0.9446	0.9456	0.9446	0.9436
GRU	0.9714	0.9727	0.9714	0.9715
BiLSTM	0.9553	0.9574	0.9553	0.9555

Table 2: Potential Assessment Predictions of Deep Hybrid Learning Model to Classify the Potato Disease using Stochastic Gradient Descent Optimizer

Decoder	Accuracy	Precision	Recall	F1 Score
LSTM	0.9928	0.9928	0.9928	0.9928
GRU	0.9946	0.9947	0.9946	0.9946
BiLSTM	0.9678	0.9678	0.9699	0.9678

Table 3: Potential Assessment Predictions of Deep Hybrid Learning Model to Classify the Potato Disease using Adam Optimizer

Decoder	Accuracy	Precision	Recall	F1 Score
LSTM	0.9446	0.9456	0.9446	0.9436
GRU	0.9714	0.9727	0.9714	0.9715
BiLSTM	0.9553	0.9574	0.9553	0.9555

Contact Information

For any questions, collaboration opportunities, or further inquiries, please feel free to reach out:

• Fatema Tuj Johora Faria

  • Email: fatema.faria142@gmail.com

• Mukaffi Bin Moin

  • Email: mukaffi28@gmail.com

Citation

@inproceedings{faria2023classification,
  title={Classification of potato disease with digital image processing technique: a hybrid deep learning framework},
  author={Faria, Fatema Tuj Johora and Moin, Mukaffi Bin and Al Wase, Ahmed and Sani, Md Rabius and Hasib, Khan Md and Alam, Mohammad Shafiul},
  booktitle={2023 IEEE 13th Annual Computing and Communication Workshop and Conference (CCWC)},
  pages={0820--0826},
  year={2023},
  organization={IEEE}
}