Plant Leaf Recognition and Classification Based on the Whale Optimization Algorithm (WOA) and Random Forest (RF)
- PDF / 957,945 Bytes
- 11 Pages / 595.276 x 790.866 pts Page_size
- 74 Downloads / 284 Views
ORIGINAL CONTRIBUTION
Plant Leaf Recognition and Classification Based on the Whale Optimization Algorithm (WOA) and Random Forest (RF) K. Pankaja1 • V. Suma2
Received: 4 July 2018 / Accepted: 11 July 2020 Ó The Institution of Engineers (India) 2020
Abstract Image processing has a vital role to play in current day scenario due to its wide band of advantages and applications such as healthcare, military, scientific and business applications. As such, plant species identification through leaf image is one of the computer vision challenges. In this paper, a method for recognizing and classifying the plant leaves by hybridizing whale optimization algorithm (WOA) and random forest (RF) is proposed. This work is carried out on Swedish and Flavia leaf datasets. Initially, pre-processing is applied to remove noises in data or to enhance its quality, prior to feature extraction. WOA is used to overcome dimensionality problem. Further, the classifier of RF is used to identify the leaf. The proposed method shows a high accuracy of 97.58% with a reduced execution time when compared with other approaches. This investigation ensures better plant leaf classification and recognition for the medical purposes. Keywords Feature extraction Feature selection Plant leaf classification Whale optimization algorithm Random forest
& K. Pankaja [email protected] V. Suma [email protected] 1
Computer Science and Engineering, Cambridge Institute of Technology, VTU, Bengaluru, India
2
Information Science and Engineering, Dayananda Sagar College of Engineering, VTU, Bengaluru, India
Introduction Plants act as an essential character in preserving the ecology of earth and environment [1]. However, many species of plants are on the verge of extinction [2]. Hence, recognition of plants plays a major role to overcome this risky situation [3]. Leaf is considered to be the most important part of a plant due to its useful characteristics for plant identification and classification [4]. Texture, color and shape of leaves are most substantial features to recognize several plants distinctly [5]. These features are beneficial in identifying and classifying a similar kind of leaves that belong to different species [6, 7]. For the purpose of classification, various plant classification methods such as k-nearest neighbor, a feed-forward backpropagation multilayered perceptron (MLP), probabilistic neural network (PNN), support vector machine (SVM) and neuro-fuzzy controller (NFC) are used. These strategies distinguish obscure plant species, in order to be perceived accurately; each species needs its own inalienable highlights. Thus, inappropriate selection of features can result in a mix of irrelevant, data redundancy that further restricts system performance. Because of higher computational complexity, these techniques are difficult. To overcome these drawbacks, feature selection technique is used to select optimal features [8, 9]. Experimentation, however, is carried out using Flavia dataset [10]. The aforementioned techniques yet had thei
Data Loading...
