agronomy

Article The Classification of Medicinal Based on Multispectral and Texture Feature Using Machine Learning Approach

Samreen Naeem 1 , Aqib Ali 1,2 , Christophe Chesneau 3 , Muhammad H. Tahir 4 , Farrukh Jamal 4 , Rehan Ahmad Khan Sherwani 5 and Mahmood Ul Hassan 6,*

1 Department of Computer Science & IT, Glim Institute of Modern Studies, Bahawalpur 63100, Pakistan; [email protected] (S.N.); [email protected] (A.A.) 2 Department of Computer Science, Concordia College Bahawalpur, Bahawalpur 63100, Pakistan 3 Department of Mathematics, Université de Caen, LMNO, Campus II, Science 3, 14032 Caen, France; [email protected] 4 Department of Statistics, The Islamia University of Bahawalpur, Bahawalpur 61300, Pakistan; [email protected] (M.H.T.); [email protected] (F.J.) 5 College of Statistical and Actuarial Sciences, University of the Punjab, Lahore 54000, Pakistan; [email protected] 6 Department of Statistics, Stockholm University, SE-106 91 Stockholm, Sweden * Correspondence: [email protected]

 Abstract: This study proposes the machine learning based classification of medical plant leaves.  The total six varieties of medicinal plant leaves-based dataset are collected from the Department of Citation: Naeem, S.; Ali, A.; Agriculture, The Islamia University of Bahawalpur, Pakistan. These are commonly named Chesneau, C.; Tahir, M.H.; Jamal, F.; in English as (herbal) Tulsi, Peppermint, Bael, Lemon balm, Catnip, and Stevia and scientifically Sherwani, R.A.K.; Ul Hassan, M. The named in Latin as Ocimum sanctum, Mentha balsamea, Aegle marmelos, Melissa officinalis, Nepeta Classification of Medicinal Plant cataria, and Stevia rebaudiana, respectively. The multispectral and digital image dataset are collected Leaves Based on Multispectral and via a computer vision laboratory setup. For the preprocessing step, we crop the region of the Texture Feature Using Machine and transform it into a gray level format. Secondly, we perform a seed intensity-based edge/line Learning Approach. Agronomy 2021, 11, 263. https://doi.org/10.3390/ detection utilizing Sobel filter and draw five regions of observations. A total of 65 fused features agronomy11020263 dataset is extracted, being a combination of texture, run-length matrix, and multi-spectral features. For the feature optimization process, we employ a chi-square feature selection approach and select Academic Editors: Thomas Scholten, 14 optimized features. Finally, five machine learning classifiers named as a multi-layer perceptron, Ruhollah Taghizadeh-Mehrjardi and logit-boost, bagging, random , and simple logistic are deployed on an optimized medicinal Karsten Schmidt plant leaves dataset, and it is observed that the multi-layer perceptron classifier shows a relatively Received: 1 January 2021 promising accuracy of 99.01% as compared to the competition. The distinct classification accuracy by Accepted: 27 January 2021 the multi-layer perceptron classifier on six medicinal plant leaves are 99.10% for Tulsi, 99.80% for Published: 30 January 2021 Peppermint, 98.40% for Bael, 99.90% for Lemon balm, 98.40% for Catnip, and 99.20% for Stevia.

Publisher’s Note: MDPI stays neutral Keywords: medicinal plant leaves; multi spectral features; texture features; classification; machine with regard to jurisdictional claims in learning; Multi-Layer Perceptron published maps and institutional affil- iations.

1. Introduction Living things on earth depend on the oxygen produced by plants. There are many Copyright: © 2021 by the authors. different types of plants, all of them playing an important role in maintaining the earth’s Licensee MDPI, Basel, Switzerland. biodiversity by providing air and water to living humans [1]. Medicinal plants are plants This article is an open access article distributed under the terms and used in the treatment and prevention of certain diseases and conditions that affect hu- conditions of the Creative Commons mans [2]. There are many different types of herbal remedies and they can vary from place to Attribution (CC BY) license (https:// place, resulting in a similar pattern of “size” and “shapes” [3]. These plants have excellent creativecommons.org/licenses/by/ medicinal properties from roots to leaves. The leaves of some such as Karpooravalli 4.0/). (Coleus ambonicus), Podina (Mentha arvensis), Neem (Adidirachta indica), Thudhuvalai

Agronomy 2021, 11, 263. https://doi.org/10.3390/agronomy11020263 https://www.mdpi.com/journal/agronomy Agronomy 2021, 11, 263 2 of 15

(Solanum trilobatum), Basil (Ocimum sanctum), etc. [4], are used in our life today. Some leaves have their own medicinal properties such as skin diseases, colds, blood purifier, indigestion [2]. Thus, medical plants are plants used for their particular properties beneficial to human health, even animal health. First called “simple” from the Middle Ages in medieval medicine, today they correspond to products from traditional or modern herbal medicine. The plant is rarely used whole; at least one of its parts (leaf, stem, root, etc.) can be used to heal itself. Different parts of the same plant can have different uses [5]. Plants with medicinal properties can also have food or condiment uses or even be used in the preparation of sanitary drinks. Since Antiquity, the theory of signatures, systematized in the 16th century, has played a major role in the distinction by analogy of plants necessary for human healing, before being extended, contested from the 17th century and completely abandoned by the elite community in the 18th century [6]. According to dissemination data, around the world, 14–28% of plants are listed as having medicinal use. Surveys carried out at the beginning of the 21st century revealed that 3–5% of patients in Western countries, 80% of rural populations in developing countries, and 85% of populations in southern Sahara use medicinal plants as their main treatment [7,8]. Nowadays, the herbal medicine market is filled with counterfeit or low-quality prod- ucts, affecting human health and sustainable development from around the world. There- fore, it becomes a hot area of research to develop tools aiming to classify herbal medicines. It is now admitted that the leaf of the plant has characteristics that are easy to extract and analyze. Therefore, it is naturally used as the main basis for the identification of all medicinal plants. With the increasing development of image processing, automatic computer image recognition is now widely used in this regard [9,10]. Image processing algorithms are used to identify the leaf images [11–20]. The back- ground behind this claim is developed below. As a first fact, medicinal plants are difficult to identify because most of them are found in deep , and the leaves look the same. If one chooses the wrong by mistake, one may have a serious health problem, which can lead to loss of human life. There are many ways to identify a plant. Plants are currently manually identified and subject to human error [11]. To avoid this, several researchers have developed an automated system identification system [12]. Many researchers work on plant leaf disease classification, segmentation, and quality assessment, for instance, Reference [13] proposed a medicinal plant classification framework using the shape and color feature of the leaf. They deployed a Support Vector Machine (SVM) classifier on the optimized features dataset and obtained 96.66% accuracy. Reference [14] proposed a plant recognition system using leaves. They used 50 medicinal images collected from google images and employed edge detection algorithms. The texture patch using Convolution Neural Networks (CNN) approach was deployed for classification and observed a 97.80% accuracy result. Reference [15] proposed a system for the classification of sugarcane leaf disease based on fungi. They used the triangle threshold approach for the segmentation of sugarcane leaves and obtained 98.60% accuracy. Reference [16] proposed a leaf image classification process using CNN. They collect 12,673 samples of the leaf of soybean and de- sign LeNet architecture and obtain a 98.32% classification result. Reference [17] presented a Romanian medicinal plant recognition framework. They used color and gray level (GL) images for his experimentation. The fused approach of Linear Discriminant Analysis (LDA) and Principal Component Analysis (PCA) are deployed and 92.9% accuracy is observed. Reference [18] suggested a novel plant disease approach utilizing a fuzzy logic-based segmentation approach. They extracted fused features that are a combination of color texture and histogram features and deployed various Machine Learning (ML) classifiers. The random forest classifier gives a very promising accuracy that is 98.4%. Reference [19] proposed the Local Binary Patterns (LBP) approach for plant classification using leaves images. They dealt with texture feature extraction, salt and pepper noise removal using ML techniques. Finally, they achieved a 93.5% cumulative classification result. Reference [20] proposed a leaf-based classification of citrus plants. They used digital images and extracted Agronomy 2021, 11, 263 3 of 15

Agronomy 2020, 10, x FOR PEER REVIEW 3 of 15 57 multi-features datasets, then optimized these features into 15 features via the ML ap- proach.images and For extracted the classification 57 multi-features purpose, datasets they employed, then optimized various these classifiers features and into it 15 has fea- been observedtures via the that ML MLP approach. gives a For higher the performanceclassification purpose, which is they 98.14% empl onoyed a region various of interestclassifi- size ofers (256 and× it 256).has been observed that MLP gives a higher performance which is 98.14% on a region of interest size of (256 × 256). Contribution Contribution The main aim of this study is to propose a framework for the classification of medicinal plantThe leaves main based aim of on this multispectral study is to propose and texture a framework features usingfor the a classification ML approach. of Thismedic- study containsinal plant six leaves steps based which on are multispectral given below: and texture features using a ML approach. This study contains six steps which are given below: • Collect multi-spectral and digital image dataset via computer vision laboratory setup. • Collect multi-spectral and digital image dataset via computer vision laboratory setup. • Crop exactly leaf region, and transform into the gray level format with (800 × 800) • Crop exactly leaf region, and transform into the gray level format with (800 × 800) resolution.resolution. • • EmployEmploy seeds seeds intensity-based intensity-based edge/lin edge/linee detection detection utilizing utilizing Sobel Sobel filter. filter. •• DrawDraw 5 5regions regions of ofobservation observation on each on each imag imagee and extract and extract fused fusedfeatures features from the from da- the taset.dataset. •• OptimizeOptimize fused fused features features dataset dataset using using chi-square chi-square feature feature selection selection approach. approach. •• ApplyApply machine machine learning learning based based classifiers classifiers for for observing observing medicinal medicinal plant plant leaves leaves classi- classifi- fication.cation.

2.2. Materials and and Methods Methods We recall that that the the medicinal medicinal plant plant leaves leaves were were collected collected from the from Department the Department of Ag- of ◦ 0 Agriculture,riculture, The The Islamia Islamia University University of Bahawalpur, of Bahawalpur, Pakistan Pakistan located located at 29°23’44” at 29 23 44”N and N and ◦ 0 7171°41’1”41 1” E [21]. [21]. The The foundation foundation of of the the dataset dataset holds holds six six types types of ofmedicinal medicinal plants plants leaves leaves commonlycommonly English named named as as (herbal) (herbal) Tulsi, Tulsi, Peppermint, Peppermint, Bael, Bael, Lemon Lemon balm, balm, Catnip, Catnip, and and SteviaStevia and and scientifically scientifically named named as (in as Latin) (in Latin) OcimumOcimum sanctum, sanctum, Mentha Mentha balsamea, balsamea, Aegle mar- Aegle marmelos,melos, Melissa Melissa officinalis, officinalis, Nepeta Nepeta cataria, cataria, and Stevia and Steviarebaudiana rebaudiana, respectively., respectively. These plants These are plants aredescribed described in Figure in Figure 1. 1.

Tulsi Peppermint Bael

Front Back Front Back Front Back

Lemon Balm Catnip Stevia

Front Back Front Back Front Back

FigureFigure 1. Sample of of the the medicinal medicinal plants plants leaves leaves belonging belonging to tothe the dataset. dataset.

InIn the experimentation, we we used used 50 50 fresh fresh leav leaveses of ofmedicinal medicinal plants plants for each for each variety. variety. Firstly,Firstly, 100 digital digital images images (50 (50 front front and and 50 50back) back) are aretaken taken for each for eachvariety. variety. So, the So, size the of size ofdigital digital image image dataset dataset of 100 of × 100 6 = ×6006 colored = 600 colored images imagesof pixel-dimension of pixel-dimension 1280 × 1024 1280 where× 1024

Agronomy 2021, 11, 263 4 of 15

Agronomy 2020, 10, x FOR PEER REVIEW 4 of 15 Agronomy 2020, 10, x FOR PEER REVIEW 4 of 15 where size of individual pixel is 0.26 mm was developed to perform further experiments. size of individual pixel is 0.26 mm was developed to perform further experiments. The size of individual pixel is 0.26 mm was developed to perform further experiments. The multiThe multispectral spectral dataset dataset was collected was collected via a mult via ai-spectral multi-spectral radiometer radiometer (MSR5) (MSR5) with 4 with feet 4 feet multi spectral dataset was collected via a multi-spectral radiometer (MSR5) with 4 feet height.height. They They extract extract five five bands bands with with range range of 460 of nm 460 to nm 1560 to 1560nm known nm known as “Red” as “Red” (R), (R), height. They extract five bands with range of 460 nm to 1560 nm known as “Red” (R), “Blue”“Blue” (B), (B), “Green” “Green” (G), (G), “Near “Near Infrared” Infrared” (NIR) (NIR) and and“Spectral “Spectral Bands Bands Shortwave Shortwave Infrared” Infrared” “Blue” (B), “Green” (G), “Near Infrared” (NIR) and “Spectral Bands Shortwave Infrared” (SWIR).(SWIR). InIn this this regard, regard, a total a total of 1200 of 1200 samples samples were were acquired, acquired, 600 samples 600 samples of digital of digital image, image, (SWIR). In this regard, a total of 1200 samples were acquired, 600 samples of digital image, andand 600 600 samples samples of ofmultispectral multispectral dataset dataset per pereach each type type of medicinal of medicinal plantplant leaves leaves utilizing utilizing and 600 samples of multispectral dataset per each type of medicinal plant leaves utilizing computercomputer vision vision laboratory laboratory setu setupp as explained as explained in Figure in Figure 2. 2. computer vision laboratory setup as explained in Figure 2.

Figure 2. Computer vision laboratory setup for medicinal plant leaves image acquisition. Figure 2. ComputerComputer vision vision laboratory laboratory setup setup for for medicinal medicinal plant plant leaves leaves image image acquisition. acquisition. The collected dataset is noise free due to the computer vision laboratory setup. For The collected collected dataset dataset is is noise noise free free due due to toth thee computer computer vision vision laboratory laboratory setup. setup. For For image pre-processing, the medicinal leaves digital image dataset is examined in OpenCV, image pre-processing, pre-processing, the the medicinal medicinal leaves leaves digital digital image image dataset dataset is examined is examined in OpenCV, in OpenCV, computer vision library [22]. Also, the Sobel filter was employed for edge/line detection computer vision vision library library [22]. [22]. Also, Also, the the Sobel Sobel filter filter was was employed employed for for edge/line edge/line detection detection as shown in Figure 3. as shown in in Figure Figure 3.3. Tulsi Peppermint Bael Tulsi Peppermint Bael

Front Back Front Back Front Back Front Back Front Back Front Back Lemon Balm Catnip Stevia Lemon Balm Catnip Stevia

Front Back Front Back Front Back Front Back Front Back Front Back Figure 3. Sample of Transformation RGB to edge/line detection process. Figure 3. SampleSample of of Transformation Transformation RGB RGB to toedge/line edge/line detection detection process. process.

Agronomy 2021, 11, 263 5 of 15

Agronomy 2020, 10, x FOR PEER REVIEW 5 of 15 × AfterAfter that,that, we crop exactly the the leaf leaf region region with with (800 (800 × 800)800) resolution, resolution, all allthe thedigital digital colorcolor imagesimages being being transformed transformed into into the the 8-bit 8-bit gray-level gray-level format, format, and draw and draw five regions five regions of ofobservation observation (ROO’s) (ROO’s) on each on each sample sample image. image. The procedure The procedure of taking of takingROOs is ROOs divided is dividedinto intotwo twosteps; steps; the first the step, first we step, take we the take size the of R sizeOOs of is ROOs(220 × 220) is (220 and,× in220) the and,second in step, the second we step,take the we size take of the ROOs size is of (280 ROOs × 280) is and (280 finally× 280) get and the different finally get datasets the different for experimenta- datasets for experimentations.tions. A total of 1000 A (5 total × 200) of 1000 ROO’s (5 ×have200) been ROO’s generated have been for each generated medicinal for plant each medicinalleaf as plantrepresented leaf as in represented Figure 4. In in this Figure manner,4. In a thistotal manner, of 6000 (6 a × total 1000) of ROO’s 6000 (6has× been1000) generated ROO’s has beenon 6 generatedvarieties of onmedical 6 varieties plant ofleaves. medical plant leaves.

Tulsi Peppermint Bael

Front Back Front Back Front Back

Lemon Balm Catnip Stevia

Front Back Front Back Front Back

FigureFigure 4.4. Sample of of transformation transformation into into gray gray level level and and draw draw 5 colorful 5 colorful regions regions of observation of observation on on medicinalmedicinal plantplant leaves dataset. dataset.

2.1.2.1. ProposedProposed Methodology The proposed methodology for the medicinal plant leaves classification is described The proposed methodology for the medicinal plant leaves classification is described be- below. In the first step, all the images acquired in the dataset were examined in OpenCV, computerlow. In the vision first step, software all the library images [ 22acquir]. Then,ed in the the Sobeldataset filter were was examined employed in OpenCV, for edge/line detection.computer Thisvision process software is basedlibrary on[22]. the Then, seed the intensity Sobel filter (pixel was threshold employed value) for edge/line of connected pixelsdetection. of an This image; process if the is thresholdbased on the value seed is intensity greater than(pixel six threshold then mark value) the of region con- called a regionnected ofpixels observation of an image; (ROO’s). if the Thethreshold graphical value representation is greater than of six the then proposed mark the methodology region forcalled the a classification region of observat of medicinalion (ROO’s). plant The leaves graphical based repr onesentation fused features of the proposed using machine learning techniques is described in Figure5.

Agronomy 2020, 10, x FOR PEER REVIEW 6 of 15

Agronomy 2021, 11, 263 6 of 15 methodology for the classification of medicinal plant leaves based on fused features us- ing machine learning techniques is described in Figure 5.

Figure 5. Proposed framework for the classification of medicinal plant leaves based on multispectral Figureand 5. textureProposed feature framework using machine for the classification learning approach. of medicinal plant leaves based on multispec- tral and texture feature using machine learning approach. 2.2. Fused Features Extraction 2.2. FusedThe Features OpenCV, Extraction computer vision software library [22] was used for the fused feature extraction process that holds texture, spectral, and gray level run length matrix features. A The OpenCV, computer vision software library [22] was used for the fused feature total of 65 fused features were extracted from each ROO’s, which is grouped as 40 texture extraction process that holds texture, spectral, and gray level run length matrix features. features, 5 multi-spectral features, and 20 run-length matrix features as described below. A total of 65 fused features were extracted from each ROO’s, which is grouped as 40 tex- The extracted dataset has a large features vector space (FVS) size of 390,000 (6000 × 65) for ture features, 5 multi-spectral features, and 20 run-length matrix features as described be- medicinal plant leaves varieties classification. low. The extracted dataset has a large features vector space (FVS) size of 390,000 (6000 × 65) for2.2.1. medicinal Texture plant Feature leaves varieties classification. The texture features are based on GL co-occurrence matrix [23–25], which is calculated via 4 dimensions (0, 45, 90, 135) degrees and distance between seeds. In this study, we used

Agronomy 2021, 11, 263 7 of 15

5 average features known as energy (ξ), inertia (τ), entropy (ψ), inverse difference (IDE), and correlation (ϕ). First, energy is defined by

2 ξ = ∑ ∑(ρuv) (1) u v

where u and v are the spatial coordinates and ρuv is gray level values. The correlation is specified by 1 ϕ = ∑ ∑(u − µu)(v − µv) ρuv (2) σuσv u v Also, the formula of the entropy is the following:

ρ ρ ψ = − ∑ ∑ uv log2 uv (3) u v

The IDE can be defined as ρ IDE = ∑ ∑ uv (4) u v |u − v|

Finally, the inertia is obtained as

2 τ = ∑ ∑(u − v) ρuv (5) u v

2.2.2. Spectral Features The frequency domain features, known as spectral features, are used in texture analysis. These features are calculated as power of different areas (A) also known as rings [26]. The numerical explanation is given as

= ( )2 Spectral Region Power ∑u∈A v∈A ∑ η u, v (6)

where η(u, v) is the frequency domain.

2.2.3. Gray Level Run-Length Matrix (GLRLM) Galloway [27,28] introduced the Gray Level Run Length Matrix (GLRM), a section of gray also known as run length. It can be described as a linear multitude of continuous pixels with the same gray level in a particular direction. The basics on this approach is recalled below. Let ηp be the number of seeds in the image, ηr be the number of discrete run lengths in the image, ψ(v1, v2|θ) be the run length matrix for an arbitrary direction θ, ηr(θ) be the number of runs in the image along angle θ and ηg be the number of discrete intensity values in the image. Then, the short run emphasis is described as

δg δr φ(w1,w2|λ) ∑w =0 ∑w =1 w 2 RLe1 = 1 2 2 (7) δr(λ)

Long run emphasis is given by

δ g δr 2 ∑w = ∑w =0 φ(w1, w2|λ)w2 RLe2 = 1 1 2 (8) δr(λ)

Gray level non-uniformity corresponds to

h i2 δg δr ∑w =1 ∑w =1 φ(w1, w2|λ) RLe3 = 1 2 (9) δr(λ) Agronomy 2021, 11, 263 8 of 15

Run percentage can be defined as

δ (λ) RLe3 = r (10) δp

Low gray level run emphasis is described as follows:

δ g δr 2 ∑w =1 ∑w =1 /w1 1 2 φ(w1, w2) RLe4 = (11) δr(λ)

High gray level run emphasis is obtained as

δ g δr 2 ∑w =1 ∑w =1 w1 1 2 φ(w1, w2) RLe5 = (12) δr(λ)

Short run low gray level emphasis is defined by

δ g δr 2 2 ∑w =1 ∑w =1 /w1w2 1 2 φ(w1, w2) RLe6 = (13) δr(λ)

Long run low gray level emphasis is determined as

δ g δr 2 2 ∑w =1 ∑w =1 w2/w1 1 2 φ(w1, w2) RLe7 = (14) δr(λ)

Finally, run length variance is presented below:

δg δr 2 RLe8 = ∑ ∑ (w2 − λ) (15) φ(w1, w2 ) w1=1 w2 =,1

2.3. Feature Selection The feature selection (FS) process is the most important part of the ML based clas- sification. This process aims to select the most valid and remove the extra features with no importance in the classification process [29]. In this study, we observe that a total of 65 fused features dataset has been extracted from each ROO’s with a large FVS size of 390,000 (6000 × 65) for medicinal plant leaves varieties classification that takes too much time in classification. The feature selection should be to identify the minimum number of columns/features from the data source that are significant in building a model [30]. Our goal in this research is to achieve better accuracy in less time. It is observed that without feature selection, the Multi-Layer Perceptron (MLP) classifier gives 98.81% accuracy results where the size of ROO is 280 × 280, but it takes a lot of time (4.83 Seconds) due to large number features. But when we go with selected features, we obtain higher accuracy in less time. There many ML based features selection approaches such as PCA technique provided excellent results on linearly separated dataset, also used in the selection of features [31]. The PCA method is an unsupervised approach [32], but the medicinal plant leaves varieties dataset is labeled, and the PCA results were not as promising on the labeled data. To solve this problem, ML based supervised feature selection techniques, namely, chi-square attribute evaluator with ranker search method [33] were used to select optimize features from the large FVS. This approach was better compared to PCA and was able to obtain the sub-dataset with the optimal characteristics for this large dataset. The chi-square attribute evaluator with ranker search method is used in ML to rank the independence of two discrete properties [34]. In FS, we specifically check whether the presence of a particular Agronomy 2021, 11, 263 9 of 15

term and the presence of a particular class is independent. Formally, when a document is given to η, we estimate the following amounts for each term and rank them according to their scores through the following formula:

(N γ γ − Eγ γ )2 Agronomy 2020, 10, x FOR PEER REVIEW 2 = l m l m 9 of 15 x(η,l,m) ∑l∈{0,1} ∑γ ∈{0;1} , (16) m Nγlγm where N is the observed frequency and E is the expected frequency, if the document con- where N is the observed frequency and E is the expected frequency, if the document tains the terms i and zero, then the value of N is 1 and if the document is in class j and contains the terms i and zero, then the value of N γ is 1 and if the document is in class zero, the value of is 1. The chi-square feature selection techniquel deployed on the me- j and zero, the value of Eγ is 1. The chi-square feature selection technique deployed on dicinal plant leaves dataset reduces lthe FVS, and gives 14 optimized features with FVS the medicinal plant leaves dataset reduces the FVS, and gives 14 optimized features with size of 84,000FVS (6000 size of× 14) 84,000 for medicinal (6000 × 14) plant for le medicinalaves classification. plant leaves Figure classification. 6 shows the Figure three-6 shows the dimensionalthree-dimensional (3D) representation (3D) of representation the optimized of features the optimized dataset featureswithin six dataset classes within using six classes PCA. The usingMDF1, PCA. MDF2, The and MDF1, MDF3 MDF2, are three and MDF3different are dimensions three different (like dimensions x, y, z) of most (like x, y, z) of discriminantmost features. discriminant features.

Figure 6. The 3D visualization of medicinal plant leaves dataset. Figure 6. The 3D visualization of medicinal plant leaves dataset. The fused optimized features for the classification of medicinal plant leaves dataset The fusedare shown optimized in Table features1. for the classification of medicinal plant leaves dataset are shown in Table 1. Table 1. Chi-square based fused optimized feature for the classification medicinal plant leaves. Table 1. Chi-square based fused optimized feature for the classification medicinal plant leaves. Sr. No. Features Sr. No. Features Sr. No. Features Sr. No. Features 1 Texture Energy Average 8 Skewness 1 Texture2 Energy Average Correlation Range 8 Skewness 9 135dgr_RLNonUni 2 3Correlation Range Inverse Diff Range9 135dgr_RLNonUni 10 R 3 4Inverse Diff Range Texture Entropy Range10 11 R G 4 Texture5 Entropy Range 45dgr_GLevNonU 11 12 G B 6 Vertl_GLevNonU 13 NIR 5 745dgr_GLevNonU S (5, 5) Entropy12 14 B SWIR 6 Vertl_GLevNonU 13 NIR 7 2.4. Classification S (5, 5) Entropy 14 SWIR Five machine learning classifiers named as multi-layer perceptron (MLP), LogitBoost 2.4. Classification (LB), Bagging (B), Random Forest (RF), and Simple Logistic (SL) are deployed on the Five machinemedicinal learning plant leaves classifiers dataset. named It is as observed multi-layer the MLPperceptron performed (MLP), well LogitBoost as compared to the (LB), Bagging (B), Random Forest (RF), and Simple Logistic (SL) are deployed on the me- dicinal plant leaves dataset. It is observed the MLP performed well as compared to the other implemented classifiers [35]. The mathematical foundations of the MLP are given below. The production of input weight and bias are summed using the summation func- tion (δ) defined by

δ = + (17)

Agronomy 2021, 11, 263 10 of 15

other implemented classifiers [35]. The mathematical foundations of the MLP are given below. The production of input weight and bias are summed using the summation function (δn) defined by k δn = ∑ ηij Ii + θi. (17) i=1

where Ii is the input variable I, k is the number of inputs, ηij is the weight, and θi is the bias Agronomy 2020, 10, x FOR PEER REVIEWterm. The activation functions of MLP is chosen as 10 of 15

where is the input variable I, k is the number of inputs,1 is the weight, and is the i(x) = (18) bias term. The activation functions of MLP is chosen1 + eas(δ n) 1 () = (18) The output of neuron j can be obtained1+e as(δ)

The output of neuron j can be obtained as k ! zi = ψi ∑ ηij Ii + θi (19) j=1 = + (19) The medicinal plant leaves classification MLP framework with all regulation parame- ters areThe shownmedicinal in Figureplant leaves7. The classification deployed MLPMLP framework classifier with with allall parametersregulation param- is defined in Tableeters are2. shown in Figure 7. The deployed MLP classifier with all parameters is defined in Table 2.

FigureFigure 7. 7. MLPMLP Based Based classification classification of medi of medicinalcinal plant plant leaves leaves framework. framework.

TableTable 2. DeployedDeployed MLP MLP classifier classifier parameters. parameters. Parameter Value Parameter Value Input Layers 1 Input LayersHidden Layers 1 14 Hidden LayersNeurons 14 18 NeuronsLearning Rate 18 0.4 Learning Rate 0.4 Momentum 0.5 Momentum 0.5 Validation Threshold 18 Validation Threshold 18 EpochsEpochs 500 500 The deployed MLP classifier with all parameters is defined in Table 2. It depends on threshold point values which are selected manually. In this study, the experiments were performedThe deployed with different MLP values classifier but, withafter alla la parametersrge number of is definedtests, these in selected Table2. Itvalues depends on thresholdbring complete point satisfaction. values which If we are increase selected or decrease manually. these In values, this study, our accuracy the experiments will be were performeddisturbed. with different values but, after a large number of tests, these selected values bring complete satisfaction. If we increase or decrease these values, our accuracy will be3. Results disturbed. and Discussion Five ML classifiers namely multi-layer perceptron (MLP), LogitBoost (LB), Bagging (B) with REPTree, Random Forest (RF), and Simple Logistic (SL), deployed on fused opti- mized features dataset for the classification of medicinal plant leaves. The foundation of the dataset holds six types of medicinal plant leaves named as Tulsi, Peppermint, Bael,

Agronomy 2021, 11, 263 11 of 15

3. Results and Discussion Five ML classifiers namely multi-layer perceptron (MLP), LogitBoost (LB), Bagging (B) with REPTree, Random Forest (RF), and Simple Logistic (SL), deployed on fused optimized

Agronomy 2020, 10, x FOR PEER REVIEWfeatures dataset for the classification of medicinal plant leaves. The foundation11 of 15 of the dataset holds six types of medicinal plant leaves named as Tulsi, Peppermint, Bael, Lemon Balm,Lemon Catnip,Balm, Catnip, and Stevia.and Stevia. The The medicinal medicinal leaves leaves classificationclassification based based on onfused fused fea- features istures performed is performed using using cross-validation cross-validation (10-fold) (10-fold) data data splitting splitting approach. approach. Different Different test- testing parametersing parameters such such as “Receiveras “Receiver Operating Operating Characteristic” Characteristic” (ROC),(ROC), “Kappa Statistics”, Statistics”, “False Positive“False Positive (FP), “Recall”(FP), “Recall” (R), “True(R), “True Positive” Positive” (TP), (TP), and and “F-Measure” “F-Measure” isis observedobserved [27]. [27]. Firstly, anFirstly, experiment an experiment performed performed on ROO’s on ROO’s size size (220 (220× ×220) 220) for for thethe classification classification of of medicinal medicinal plant plant leaves and observed a well-organized accuracy which is 95.87 %, 95.04 %, 94.21%, leaves and observed a well-organized accuracy which is 95.87%, 95.04%, 94.21%, 93.38%, 93.38 %, and 92.56% using MLP, LB, B, RF, and SL, respectively, as shown in Table 3. and 92.56% using MLP, LB, B, RF, and SL, respectively, as shown in Table3. Table 3. Classification of medicinal plant leaves using five ML classifiers with ROO’s size of (220 × Table 3. Classification220). of medicinal plant leaves using five ML classifiers with ROO’s size of (220 × 220).

Classifiers Kappa Statistics TP RateKappa FP Sta- RateTP Recall F-Measure ROCTime (Sec) Precision Classifiers FP Rate Recall F-Measure ROC Precision tistics Rate (Sec) MLP 0.9504 0.959 0.008 0.959 0.958 0.999 0.19 0.961 LB 0.9405MLP 0.950 0.9504 0.010 0.959 0.9500.008 0.959 0.950 0.958 0.9890.999 0.19 0.11 0.961 0.951 B 0.9306LB 0.942 0.9405 0.012 0.950 0.9420.010 0.950 0.941 0.950 0.9910.989 0.11 0.3 0.951 0.944 SLg 0.9207B 0.934 0.9306 0.013 0.942 0.9340.012 0.942 0.934 0.941 0.9600.991 0.3 0.10 0.944 0.935 RF 0.9107SLg 0.926 0.9207 0.015 0.934 0.9260.013 0.934 0.926 0.934 0.9550.960 0.10 0.7 0.935 0.927 RF 0.9107 0.926 0.015 0.926 0.926 0.955 0.7 0.927 It is observed that the MLP performs efficiently insisted of other employed classifiers It is observed that the MLP performs efficiently insisted of other employed classifiers × whenwhen the size size of of ROO’s ROO’s 220 220 × 220,220, as shown as shown in Figure in Figure 8. 8 .

RF SLg B LB MLP

80.00% 85.00% 90.00% 95.00% 100.00% MLP LB B SLg RF Accuracy 95.87% 95.04% 94.21% 93.38% 92.56%

FigureFigure 8. MedicinalMedicinal plant plant leaves leaves classifica classificationtion graphic graphic on ROI’s on ROI’s(220 × (220200). × 200).

For improvement in in classification classification result, result, the the proposed proposed approach approach employed employed on me- on medici- naldicinal plant plant leaves leaves dataset dataset where where the the size size ofof ROO’sROO’s is is (280 (280 ×× 280).280). We We observe observe very very prom- promising ising results which are 99.01 %, 98.01 %, 97.02%, 96.03%, and 95.04% using MLP, LB, B, results which are 99.01%, 98.01%, 97.02%, 96.03%, and 95.04% using MLP, LB, B, RF, and RF, and SL, respectively, as shown in Table 4. SL, respectively, as shown in Table4. Table 4. Classification of medicinal plant leaves using five ML classifiers with ROO’s size of (280 × Table 4. Classification280). of medicinal plant leaves using five ML classifiers with ROO’s size of (280 × 280).

Classifiers Kappa Statistics TP RateKappa FP Sta- RateTP Recall F-Measure ROCTime (Sec) Precision Classifiers FP Rate Recall F-Measure ROC Precision tistics Rate (Sec) MLP 0.9876 0.990 0.002 0.990 0.990 0.998 0.13 0.991 LogitBoost 0.9752MLP 0.980 0.9876 0.005 0.990 0.9800.002 0.990 0.981 0.990 0.9990.998 0.13 0.19 0.991 0.981 Bagging 0.9629LogitBoost 0.970 0.9752 0.007 0.980 0.9700.005 0.980 0.971 0.981 0.995 0.999 0.19 0.11 0.981 0.974 SLg 0.9506Bagging 0.960 0.9629 0.007 0.970 0.9600.007 0.970 0.965 0.971 0.9840.995 0.11 0.13 0.974 0.970 RF 0.9381SLg 0.950 0.9506 0.013 0.960 0.9500.007 0.960 0.951 0.965 0.9850.984 0.13 0.9 0.970 0.956 RF 0.9381 0.950 0.013 0.950 0.951 0.985 0.9 0.956

It is observed that MLP effectively emphasizes the other classifiers when the size of ROO is 280 × 280, as shown in Figure 9.

Agronomy 2021, 11, 263 12 of 15

It is observed that MLP effectively emphasizes the other classifiers when the size of Agronomy 2020, 10, x FOR PEER REVIEWROO is 280 × 280, as shown in Figure9. 12 of 15 Agronomy 2020, 10, x FOR PEER REVIEW 12 of 15

RF RF SLg SLg B B LB LB MLP MLP 80.00% 85.00% 90.00% 95.00% 100.00% 80.00% 85.00% 90.00% 95.00% 100.00% MLP LB B SLg RF Accuracy 99.01%MLP 98.01% LB 97.02% B 96.03% SLg 95.04% RF Accuracy 99.01% 98.01% 97.02% 96.03% 95.04%

Figure 9. Medicinal plant leaves classification graphic on ROI’s (280 × 280). FigureFigure 9. 9.Medicinal Medicinal plantplant leaves leaves classification classification graphic graphic on on ROI’s ROI’s (280 (280× × 280).280). The Confusion Matrix (CM) of MLP classifier on fused medicinal plant leaves dataset usingTheThe ROO’s Confusion Confusion size 280 Matrix Matrix × 280. (CM) (CM) The of diagonalof MLP MLP classifier classifier cells contain on on fused fused the medicinal numbermedicinal of plant plantcorrectly leaves leaves identified dataset dataset usingplantsusing ROO’s ROO’s and the size size rest 280 280 of× the× 280.280. cells TheThe contain diagonaldiagonal the cellsnumbercells containcontain of misclassified thethe numbernumber plants. ofof correctlycorrectly This identifiedCM identified is pre- plantssentedplants and andin Table thethe rest 5. of the the cells cells contain contain the the number number of of misclassified misclassified plants. plants. This This CM CM is pre- is presentedsented in inTable Table 5.5 . Table 5. CM showing medicinal plant leaves classification on ROO’s size (280 × 280) using MLP. Table 5. CM showingTable medicinal5. CM showing plant leavesmedicinal classification plant leaves on classification ROO’s size (280 on ×ROO’s280) usingsize (280 MLP. × 280) using MLP. Accu- Classes Tulsi PeppermintClasses BaelTulsi LemonPeppermint Balm Bael Catnip Lemon Balm Stevia Catnip TotalStevia Total Accuracy Accu- Classes Tulsi Peppermint Bael Lemon Balm Catnip Stevia Total racy Tulsi 991 1Tulsi 2991 01 2 60 06 10000 1000 99.1% 99.1%racy Peppermint 0 988 0 2 5 5 1000 98.8% PeppermintTulsi 9910 9881 02 20 56 50 1000 98.8%99.1% Bael 4 6 984 0 3 3 1000 98.4% Peppermint 0 988 0 2 5 5 1000 98.8% Lemon Bael 4 6 984 0 3 3 1000 98.4% 0 1 0 999 0 0 1000 99.9% Balm LemonBael Balm 04 16 9840 9990 03 03 1000 99.9%98.4% Catnip 0 4LemonCatnip Balm 0 0 041 0 9949990 29940 100020 1000 99.4% 99.4%99.9% Stevia 3 0 2 3 0 992 1000 99.2% CatnipStevia 30 04 20 30 9940 9922 1000 99.2%99.4% Stevia 3 0 2 3 0 992 1000 99.2% TheThe distinctdistinct classification classification accuracy accuracy of of six six medicinal medicinal plant plant leaves, leaves, named named as Tulsi, as Tulsi, Pep- Peppermint,permint,The distinctBael, Bael, Lemon classification Lemon balm, balm, Catnip, accuracy Catnip, and andof sixStev Stevia, medicinalia, were were plant99.10%, 99.10%, leaves, 99.80%, named 98.40%,98.40%, as Tulsi, 99.90%,99.90%, Pep- × 99.40%,99.40%,permint, and and Bael, 99.20%, 99.20%, Lemon respectively, respectively, balm, Catnip, on on ROO’s ROO’s and Stev size sizeia, 280 280 were × 280280 99.10%, asas shownshown 99.80%, inin FigureFigure 98.40%, 10 10.. 99.90%, 99.40%, and 99.20%, respectively, on ROO’s size 280 × 280 as shown in Figure 10. 100.00% 99.90% 100.00% 99.90% 99.40% 99.50% 99.10% 99.40% 99.20% 99.50% 99.20% 99.00% 99.10% 98.80% 99.00% 98.80% 98.40% 98.50% 98.40% 98.50% 98.00% 98.00% 97.50% 97.50% Tulsi Peppermint Bael Lemon Balm Catnip Stevia Tulsi Peppermint Bael Lemon Balm Catnip Stevia Tulsi Peppermint Bael Lemon Balm Catnip Stevia Tulsi Peppermint Bael Lemon Balm Catnip Stevia

FigureFigure 10. 10.MLP MLP basedbased classification classification of of six six medicinal medicinal plant plant leaf leaf varieties varieties on on ROO’s ROO’s size size (250 (250× 250).× Figure250). 10. MLP based classification of six medicinal plant leaf varieties on ROO’s size (250 × 250). We started our experimentations with the size of ROOs 220 × 220. After that, we grad- uallyWe increased started theour sizeexperimentations of ROOs to achieve with the better size ofaccuracy. ROOs 220 Finally, × 220. atAfter the that,size weof ROOs grad- ually increased the size of ROOs to achieve better accuracy. Finally, at the size of ROOs

Agronomy 2021, 11, 263 13 of 15

We started our experimentations with the size of ROOs 220 × 220. After that, we Agronomy 2020, 10, x FOR PEER REVIEW 13 of 15 gradually increased the size of ROOs to achieve better accuracy. Finally, at the size of × ROOs280 × 280280, we280, observe we observe the promising the promising accuracy accuracy because because it covers it covers maximum maximum useful useful infor- information.mation. Further Further increase increase in the in size the sizeof ROOs of ROOs was wascausing causing a decrease a decrease in the in theaccuracy accuracy due dueto speckle to speckle noise. noise. Lastly, Lastly, the comparative the comparative analys analysisis performs performs for the for classification the classification of medic- of medicinal plant leaves with the sizes of ROO’s 220 × 220 and 280 × 280, respectively, as inal plant leaves with the sizes of ROO’s 220 × 220 and 280 × 280, respectively, as shown shown in Figure 11. in Figure 11.

100.00% 98.00% 96.00% 94.00% 92.00% 90.00% 88.00% MLP LB B SLg RF ROO's (220 × 220) 95.87% 95.04% 94.21% 93.38% 92.56% ROO's (280 × 280) 99.01% 98.01% 97.02% 96.03% 95.04%

FigureFigure 11.11. Comparative analysis for for the the classification classification of of medicinal medicinal plant plant leaves leaves using using ML ML based based classifiers where ROO’s Size (220 × 220) and (280 × 280). classifiers where ROO’s Size (220 × 220) and (280 × 280).

TheThe methodology methodology proposed proposed is comparativelyis comparativel reliabley reliable and efficientand efficient from thatfrom described that de- previouslyscribed previously [13,14,16 –[13,14,16–20].20]. Furthermore, Furthermore, it is consistent, it is consistent, satisfactory, satisfactory, and better and from better the existingfrom the medicinal existing medicinal plant leaves plant classification. leaves classification. A comparative A comparative analysis analysis of the proposedof the pro- methodologyposed methodology with existing with existing works isworks shown is inshown Table in6. Table 6.

TableTable 6. 6.Comparison Comparison of of proposed proposed methodology methodology with with existing existing methodologies. methodologies.

ReferenceReference FeaturesFeatures Classifiers Classifiers Accuracy Accuracy [13] Shape and Color Features SVM 96.66% [13] Shape and Color Features SVM 96.66% [14[14]] TextureTexture Features Features CNN CNN 97.80% 97.80% [16[16]] Morphological Morphological Features Features CNN, CNN, LeNet LeNet 98.32% 98.32% [17[17]] Texture Texture Features Features PCA, PCA, LDA LDA 92.90% 92.90% [18[18]] FusedFused Features Features RF RF 98.40% 98.40% [19] Texture Features LBP 93.50% [20[19]] MultiTexture Features Features LBP MLP 93.50% 98.14% [20] Multi Features MLP 98.14% Proposed Proposed Meth-MultiMulti Spectral Spectral + Texture + Texture Features Fea- MLP 99.01% Methodology MLP 99.01% odology tures 4. Conclusions 4. Conclusions In this study, we develop a machine learning (ML) based medical plants leaves classificationIn this study, utilizing we multispectraldevelop a machine and texture learning dataset. (ML) Thebased main medical objective plants is toleaves collect clas- a refinedsification and utilizing standardized multispectral dataset, edge/lineand texture detection, dataset. fusedThe main features objective extraction, is to collect optimized a re- extractedfined and features, standardized and select dataset, the most edge/line valuable dete featurection, andfused select features the efficient extraction, ML classifiers.optimized Theextracted fused features, (multispectral and select + texture) the most feature valuable dataset feature holds and six select types the of efficient medicinal ML leaves classi- namedfiers. The Tulsi, fused Peppermint, (multispectral Bael, + Lemontexture)Balm, feature Catnip, dataset and holds Stevia six types collected of medicinal via computer leaves visionnamed laboratory Tulsi, Peppermint, setup. Due Bael, to Lemon the complex Balm, laboratoryCatnip, and setup, Stevia the collected collected via dataset computer is veryvision refined laboratory and standardized.setup. Due to the The complex chi-square laboratory feature setup, selection the approachcollected dataset provides is very the 14refined most worthfuland standardized. features that The are chi-square useful to obtainfeature better selection classification approach results. provides A totalthe 14 of most five AI-basedworthful classifiersfeatures that are are considered, useful to namedobtain asbetter multi-layer classification perceptron results. (MLP), A total LogitBoost of five AI- (LB),based Bagging classifiers (B), are Random considered, Forest named (RF), and as mu Simplelti-layer Logistic perceptron (SL). Firstly, (MLP), an LogitBoost experiment (LB), is Bagging (B), Random Forest (RF), and Simple Logistic (SL). Firstly, an experiment is per- formed on ROO’s size (220 × 220) for the classification of medicinal plant leaves dataset. It is obtained a well-organized accuracy which are 95.87%, 95.04%, 94.21%, 93.38%, and 92.56% for MLP, LB, B, RF, and SL, respectively. Secondly, the same approach is employed

Agronomy 2021, 11, 263 14 of 15

performed on ROO’s size (220 × 220) for the classification of medicinal plant leaves dataset. It is obtained a well-organized accuracy which are 95.87%, 95.04%, 94.21%, 93.38%, and 92.56% for MLP, LB, B, RF, and SL, respectively. Secondly, the same approach is employed on a medicinal plant leaves dataset where the size of ROO’s is (280 × 280). We obtain very promising results which are 99.01%, 98.01%, 97.02%, 96.03%, and 95.04% respectively. In addition, we observe that the MLP classifier performed well as compared to other implemented AI-based classifiers. This study opens a new horizon in the field of medicinal plant leaves classification. Also, it can be very helpful for pharmacists to recognize the correct medical plant and will help in the process of making medicine.

Limitation and Future Works This study is limited to six medicinal plant leaves while there are millions of types of medicinal plant/herbs in the world. This is a pixel-based approach and in the future, we may wish to use an object-based approach. In future this proposed approach deployed on other medicinal plant leaves also proposed approach can be improved using hyper spectral and 3D digital image dataset.

Author Contributions: S.N., Data curation; A.A., Conceptualization, Methodology, Writing—original draft, Software; C.C., Writing—review and editing; M.H.T., Project administration; F.J., Validation, Formal analysis, Writing—review and editing; R.A.K.S., Supervision; M.U.H., Resources, Investiga- tion; All authors have read and agreed to the published version of the manuscript. Funding: This research received no external funding. Data Availability Statement: Data available on request due to restrictions eg privacy or ethical. The data presented in this study are available on request from the corresponding author. The data are not publicly available due to it is self-generated dataset. Acknowledgments: The authors thank anonymous referees for careful reading of the manuscript and constructive comments, that significantly improved this paper. Samreen Naeem and Aqib Ali are thankful to their master supervisor, Salman Qadri, Department of Information Technology, The Islamia University of Bahawalpur, Pakistan for his motivational support. Conflicts of Interest: The authors declare no conflict of interest.

References 1. Nawkar, M.G.; Maibam, P.; Park, J.H.; Sahi, V.P.; Lee, S.Y.; Kang, C.H. UV-induced cell death in plants. Int. J. Mol. Sci. 2013, 14, 1608–1628. [CrossRef][PubMed] 2. Salehi, B.; Kumar, N.V.; ¸Sener, B.; Sharifi-Rad, M.; Kılıç, M.; Mahady, G.B.; Vlaisavljevic, S.; Iriti, M.; Kobarfard, F.; Setzer, W.N.; et al. Medicinal plants used in the treatment of human immunodeficiency virus. Int. J. Mol. Sci. 2018, 19, 1459. [CrossRef] [PubMed] 3. Nelly, A.; Annick, D.D.; Frederic, D. Plants used as remedies antirheumatic and antineuralgic in the traditional medicine of Lebanon. J. Ethnopharmacol. 2008, 120, 315–334. 4. Raskin, I.; Ribnicky, D.M.; Komarnytsky, S.; Ilic, N.; Poulev, A.; Borisjuk, N.; Brinker, A.; Moreno, D.A.; Ripoll, C.; Yakoby, N.; et al. Plants and human health in the twenty-first century. Trends Biotechnol. 2002, 20, 522–531. [CrossRef] 5. Leonti, M. The future is written: Impact of scripts on the cognition, selection, knowledge and transmission of medicinal plant use and its implications for ethnobotany and ethnopharmacology. J. Ethnopharmacol. 2011, 134, 542–555. [CrossRef] 6. Mamedov, N. Medicinal plants studies: History, challenges and prospective. Med. Aromat. Plants 2012, 1, e133. [CrossRef] 7. Amenu, E. Use and Management of Medicinal Plants by Indigenous People of Ejaji Area (Chelya Woreda) West Shoa, Ethiopia: An Ethnobotanical Approach. Master’s Thesis, University in Addis, Ababa, Ethiopia, 2007. 8. Hu, R.; Lin, C.; Xu, W.; Liu, Y.; Long, C. Ethnobotanical study on medicinal plants used by Mulam people in Guangxi, China. J. Ethnobiol. Ethnomed. 2020, 16, 1–50. [CrossRef] 9. Pferschy-Wenzig, E.M.; Bauer, R. The relevance of pharmacognosy in pharmacological research on herbal medicinal products. Epilepsy Behav. 2015, 52, 344–362. [CrossRef] 10. Oppong, P.K. The Influence of Packaging and Brand Equity on Over-the-Counter Herbal Medicines in Kumasi, Ghana. Ph.D. Thesis, University of KwaZulu-Natal, Durban, South African, 2018. [CrossRef] 11. Zhang, F.; Zhang, X. Classification and quality evaluation of tobacco leaves based on image processing and fuzzy comprehensive evaluation. Sensors 2011, 11, 2369–2384. [CrossRef] Agronomy 2021, 11, 263 15 of 15

12. Rahmatullah, M.; Azam, M.N.; Rahman, M.M.; Seraj, S.; Mahal, M.J.; Mou, S.M.; Nasrin, D.; Khatun, Z.; Islam, F.; Chowdhury, M.H. Chowdhury. A survey of medicinal plants used by Garo and non-Garo traditional medicinal practitioners in two villages of Tangail district, . Am. Eurasian J. Sustain. Agric. 2011, 5, 350–357. 13. Dahigaonkar, T.D.; Kalyane, R. Identification of ayurvedic medicinal plants by image processing of leaf samples. Int. Res. J. Eng. Technol. (Irjet) 2018, 5, 351–355. 14. Sabarinathan, C.; Hota, A.; Raj, A.; Dubey, V.K.; Ethirajulu, V. Medicinal plant leaf recognition and show medicinal uses using convolutional neural network. Int. J. Glob. Eng. 2018, 1, 120–127. 15. Khirade, S.D.; Patil, A.B. Plant disease detection using image processing. In Proceedings of the 2015 International Conference on Computing Communication Control and Automation, Pune, , 26–27 February 2015; pp. 768–771. 16. Wallelign, S.; Polceanu, M.; Buche, C. Soybean plant disease identification using convolutional neural network. In Proceedings of the Thirty-First International Flairs Conference, Melbourne, FL, USA, 10 May 2018. 17. Simion, I.M.; Casoni, D.; Sârbu, C. Classification of Romanian medicinal plant extracts according to the therapeutic effects using thin layer chromatography and robust chemometrics. J. Pharm. Biomed. Anal. 2019, 163, 137–143. [CrossRef][PubMed] 18. Dhingra, G.; Kumar, V.; Joshi, H.D. A novel computer vision based neutrosophic approach for leaf disease identification and classification. Measurement 2019, 135, 782–794. [CrossRef] 19. Turkoglu, M.; Hanbay, D. Leaf-based plant species recognition based on improved local binary pattern and extreme learning machine. Phys. A Stat. Mech. Appl. 2019, 527, 121297. [CrossRef] 20. Qadri, S.; Furqan Qadri, S.; Husnain, M.; Saad Missen, M.M.; Khan, D.M.; Muzammil-Ul-Rehman Razzaq, A.; Ullah, S. Machine vision approach for classification of citrus leaves using fused features. Int. J. Food Prop. 2019, 22, 2072–2089. [CrossRef] 21. The Islamia University of Bahawalpur, Pakistan. 2020. Available online: https://www.iub.edu.pk/faculty-of-agriculture-and- environmental-sciences?1=1forAgriculturalfarms (accessed on 1 October 2020). 22. Bradski, G.; Kaehler, A. Learning OpenCV: Computer Vision with the OpenCV Library; O’Reilly Media, Inc.: Newton, MA, USA, 2008. 23. Galloway, M. Texture analysis using gray level run lengths. Comput. Graph. Image Process 1975, 4, 172–179. [CrossRef] 24. Naeem, S.; Ali, A.; Qadri, S.; Mashwani, W.K.; Tairan, N.; Shah, H.; Fayaz, M.; Jamal, F.; Chesneau, C.; Anam, S. Machine-Learning Based Hybrid-Feature Analysis for Liver Cancer Classification Using Fused (MR and CT) Images. Appl. Sci. 2020, 10, 3134. [CrossRef] 25. Haralick, R.M.; Shanmugam, K.; Dinstein, I.H. Textural features for image classification. IEEE Trans. Syst. ManCybern. 1973, 6, 610–621. [CrossRef] 26. Bantan, R.A.; Ali, A.; Naeem, S.; Jamal, F.; Elgarhy, M.; Chesneau, C. Discrimination of sunflower seeds using multispectral and texture dataset in combination with region selection and supervised classification methods. Chaos Interdiscip. J. Nonlinear Sci. 2020, 30, 113142. [CrossRef] 27. Ali, A.; Qadri, S.; Khan Mashwani, W.; Kumam, W.; Kumam, P.; Naeem, S.; Goktas, A.; Jamal, F.; Chesneau, C.; Anam, S.; et al. Machine Learning Based Automated Segmentation and Hybrid Feature Analysis for Diabetic Retinopathy Classification Using Fundus Image. Entropy 2020, 22, 567. [CrossRef][PubMed] 28. Abbas, Z.; Rehman, M.; Najam, S.; Rizvi, S.M.D. An e_cient gray-level co-occurrence matrix (GLCM) based approach towards classification of skin lesion. In Proceedings of the 2019 Amity International Conference on Artificial Intelligence (AICAI), Dubai, UAE, 4–6 February 2019; pp. 317–320. 29. Ali, A.; Mashwani, W.K.; Tahir, M.H.; Belhaouari, S.B.; Alrabaiah, H.; Naeem, S.; Nasir, J.A.; Jamal, F.; Chesneau, C. Statistical features analysis and discrimination of maize seeds utilizing machine vision approach. J. Intell. Fuzzy Syst. 2021, 40, 703–714. [CrossRef] 30. Behura, A. The Cluster Analysis and Feature Selection: Perspective of Machine Learning and Image Processing. Data Anal. Bioinf. A Mach. Learn. Perspect. 2021, 249–280. [CrossRef] 31. Cardinali, F.; Bracciale, M.P.; Santarelli, M.L.; Marrocchi, A. Principal Component Analysis (PCA) Combined with Naturally Occurring Crystallization Inhibitors: An Integrated Strategy for a more Sustainable Control of Salt Decay in Built Heritage. Heritage 2021, 4, 13. [CrossRef] 32. Ali, A.; Qadri, S.; Mashwani, W.K.; Brahim Belhaouari, S.; Naeem, S.; Rafique, S.; Jamal, F.; Chesneau, C.; Anam, S. Machine learning approach for the classification of corn seed using hybrid features. Int. J. Food Prop. 2020, 23, 1110–1124. [CrossRef] 33. Gnanambal, S.; Thangaraj, M.; Meenatchi, V.T.; Gayathri, V. Classification algorithms with attribute selection: An evaluation study using weka. Int. J. Adv. Netw. Appl. 2018, 9, 3640–3644. 34. McHugh, M.L. The chi-square test of independence. Biochem. Med. 2013, 23, 143–149. [CrossRef] 35. Ali, A.; Nasir, J.A.; Ahmed, M.M.; Naeem, S.; Anam, S.; Jamal, F.; Chesneau, C.; Zubair, M.; Anees, M.S. Machine Learning Based Statistical Analysis of Emotion Recognition using Facial Expression. RADS J. Biol. Res. Appl. Sci. 2020, 11, 39–46. [CrossRef]