We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
Capsule network-based disease classification for Vitis Vinifera leaves.
- Authors
Andrushia, A. Diana; Neebha, T. Mary; Patricia, A. Trephena; Sagayam, K. Martin; Pramanik, Sabyasachi
- Abstract
The primary source of food is extracted from the plant. Take care of and maintain the plants in real-time to enhance human survival. Diseases in plants can directly lead to a reduction in the industrial economy. Automatic detection of plant diseases is crucial for effective disease control. Vitis vinifera (Grapes) is an essential crop with rich vitamin C nutrients. This paper targets to classify the major diseases in Vitis vinifera leaves using Capsule Network (CapsNet), that prevails over the significant CNN limitations by discarding the pooling layers and adding capsule layers. Dynamic routing techniques of CapsNet make are more robust for the affine transformation of the leaves dataset. It is capable of learning large datasets effectively with vital image transformations such as rotations and transitions. Implementing CapsNet for Vitis vinifera leaf disease classification, which utilizes dynamic routing between capsules, is a novel method. The proposed CapsNet for disease classification is trained with augmented and non-augmented datasets. The performance metrics highlight that the proposed method can effectively classify the Vitis vinifera plant leaves with 98.7% of validation accuracy. The results highlight that the proposed model performs well in detecting and classifying the diseases of Vitis vinifera plant leaves on two benchmark datasets.
- Subjects
NOSOLOGY; CAPSULE neural networks; AFFINE transformations; VITAMIN C; GRAPES; VITIS vinifera; PLANT diseases
- Publication
Neural Computing & Applications, 2024, Vol 36, Issue 2, p757
- ISSN
0941-0643
- Publication type
Article
- DOI
10.1007/s00521-023-09058-y