Esraa Elhariri

Fruit-Based Tomato Grading System Using Features Fusion and Support Vector Machine

Machine learning and computer vision techniques have applied for evaluating food quality as well as crops grading. In this paper, a new classification system has been proposed to classify infected/uninfected tomato fruits according to its external surface. The system is based on feature fusion method with color and texture features. Color moments, GLCM, and Wavelets energy and entropy have been used in the proposed system. Principle Component Analysis (PCA) technique has been used to reduce the feature vector obtained after fusion to avoid dimensionality problem and save time and cost. Support vector machine (SVM) was used to classify tomato images into 2 classes; infected/uninfected using Min-Max and Z-Score normalization methods. The dataset used in this research contains 177 tomato fruits each was captured from four faces (Top, Side1, Side2, and End). Using 70% of the total images for training phase and 30% for testing, our proposed system achieved accuracy 92%.

Multi-class SVM Based Classification Approach for Tomato Ripeness

This article presents a content-based image classification system to monitor the ripeness process of tomato via investigating and classifying the different maturity/ripeness stages. The proposed approach consists of three phases; namely pre-processing, feature extraction, and classification phases. Since tomato surface color is the most important characteristic to observe ripeness, this system uses colored histogram for classifying ripeness stage. It implements Principal Components Analysis (PCA) along with Support Vector Machine (SVM) algorithms for feature extraction and classification of ripeness stages, respectively. The datasets used for experiments were constructed based on real sample images for tomato at different stages, which were collected from a farm at Minia city. Datasets of 175 images and 55 images were used as training and testing datasets, respectively. Training dataset is divided into 5 classes representing the different stages of tomato ripeness. Experimental results showed that the proposed classification approach has obtained ripeness classification accuracy of 92.72%, using SVM linear kernel function with 35 images per class for training.

RoadMonitor: An Intelligent Road Surface Condition Monitoring System

Well maintained road network is an essential requirement for the safety and consistency of vehicles moving on that road and the wellbeing of people in those vehicles. On the other hand, guaranteeing an adequate maintenance by road managers can be achieved via having sufficient and accurate information concerning road infrastructure quality that can be as well utilized concurrently by the widespread means of users’ mobile devices both locally and worldwide. This article proposes a road condition monitoring framework that detects the road anomalies such as speed bumps. In the proposed approach, the main indicator for road anomalies is the gyroscope around gravity rotation in addition to the accelerometer sensor as a cross-validation method to confirm the detection results that were gathered from the gyroscope.

Plant classification system based on leaf features

This paper presents a classification approach based on Random Forests (RF) and Linear Discriminant Analysis (LDA) algorithms for classifying the different types of plants. The proposed approach consists of three phases that are pre-processing, feature extraction, and classification phases. Since most types of plants have unique leaves, so the classification approach presented in this research depends on plants leave. Leaves are different from each other by characteristics such as the shape, color, texture and the margin. The used dataset for this experiments is a database of different plant species with total of only 340 leaf images, was downloaded from UCI- Machine Learning Repository. It was used for both training and testing datasets with 10-fold cross-validation. Experimental results showed that LDA achieved classification accuracy of (92.65%) against the RF that achieved accuracy of (88.82%) with combination of shape, first order texture, Gray Level Co-occurrence Matrix (GLCM), HSV color moments, and vein features.

Grey wolf optimization for one-against-one multi-class support vector machines

Grey Wolf Optimization (GWO) algorithm is a new meta-heuristic method, which is inspired by grey wolves, to mimic the hierarchy of leadership and grey wolves hunting mechanism in nature. This paper presents a hybrid model that employs grey wolf optimizer (GWO) along with support vector machines (SVMs) classification algorithm to improve the classification accuracy via selecting the optimal settings of SVMs parameters. The proposed approach consists of three phases; namely pre-processing, feature extraction, and GWO-SVMs classification phases. The proposed classification approach was implemented by applying resizing, remove background, and extracting color components for each image. Then, feature vector generation has been implemented via applying PCA feature extraction. Finally, GWO-SVMs model is developed for selecting the optimal SVMs parameters. The proposed approach has been implemented via applying One-againstOne multi-class SVMs system using 3-fold cross-validation. The datasets used for experiments were constructed based on real sample images of bell pepper at different stages, which were collected from farms in Minya city, Upper Egypt. Datasets of total 175 images were used for both training and testing datasets. Experimental results indicated that the proposed GWO-SVMs approach achieved better classification accuracy compared to the typical SVMs classification algorithm.

Machine Learning Based Classification Approach for Predicting Students Performance in Blended Learning

Nowadays, recognizing and predicting students learning achievement introduces a significant challenge, especially in blended learning environments, where online (web-based electronic interaction) and offline (direct face-to-face interaction in classrooms) learning are combined. This paper presents a Machine Learning (ML) based classification approach for students learning achievement behavior in Higher Education. In the proposed approach, Random Forests (RF) and Support Vector Machines (SVM) classification algorithms are being applied for developing prediction models in order to discover the underlying relationship between students past course interactions with Learning Management Systems (LMS) and their tendency to pass/fail. In this paper, we considered daily students interaction events, based on time series, with a number of Moodle LMS modules as the leading characteristics to observe students learning performance. The dataset used for experiments is constructed based on anonymized real data samples traced from web-log files of students access behavior concerning different modules in a Moodle online LMS throughout two academic years. Experimental results showed that the proposed RF classification system has outperformed the typical SVMs classification algorithm.

A Hybrid Classification Model for EMG Signals Using Grey Wolf Optimizer and SVMs

Electromyography (EMG) signal is an electrical indicator for neuromuscular activation. It provides direct access to physiological processes enabling the muscle to generate force and produce movement in order to accomplish countless functions. As a successful classification of the EMG signal is basically dependent on the selection of the best parameters carefully, this paper proposes a hybrid optimized classification model for EMG signals classification. The proposed system implements grey wolf optimizer (GWO) combined with support vector machines (SVMs) classification algorithm in order to improve the classification accuracy via selecting the optimal settings of SVMs parameters. The proposed approach consists of three phases; namely pre-processing, feature extraction, and GWO-SVMs classification phases. The obtained experimental results obviously indicate that significant enhancements in terms of classification accuracy have been achieved by the proposed GWO-SVMs classification system. It has outperformed the typical SVMs classification algorithm via achieving an accuracy of over 90 % using the radial basis function (RBF) kernel function.

Random Forests Based Classification for Crops Ripeness Stages

This article presents a classification approach based on random forests algorithm for estimating and classifying the different maturity/ripeness stages of two types of crops; namely tomato and bell pepper (sweet pepper). The proposed approach consists of three phases that are pre-processing, feature extraction, and classification phases. Surface color of tomato and bell pepper is the most important characteristic to observe ripeness. So, the proposed classification system uses color features for classifying ripeness stages. It implements principal components analysis (PCA) along with support vector machine (SVM) algorithms and random forests (RF) classifier for features extraction and classification of ripeness stages, respectively. The datasets used for experiments were constructed based on real sample images for both tomatoes and bell pepper at different stages, which were collected from farms in Minya city, Upper Egypt. Datasets of total 250 and 175 images for tomato and bell pepper, respectively were used for both training and testing datasets. Training dataset is divided into five classes representing the different stages of tomato and bell pepper ripeness. Experimental results showed that SVM with Linear Kernel function achieved accuracy better than RF.

Bio-inspired optimization for feature set dimensionality reduction

In this paper, two novel bio-inspired optimization algorithms; namely Dragonfly Algorithm (DA) and Grey Wolf Optimizer (GWO), have been applied for fulfilling the goal of feature set dimensional reduction. The proposed classification system has been tested via solving the problem of Electromyography (EMG) signal classification with optimal features subset selection. The obtained experimental results showed that the GWO based Support Vector Machines (SVM) classification algorithm has achieved an accuracy of 93.22% using 31% of the total extracted features. It also outperformed both the typical SVM algorithm, with no feature set optimization, and the DA based optimized feature set SVM classification, for the tested EMG dataset.

Cultivation-time Recommender System Based on Climatic Conditions for Newly Reclaimed Lands in Egypt

This research proposes cultivation-time recommender system for predicting the best sowing dates for winter cereal crops in the newly reclaimed lands in Farafra Oasis, The Egyptian Western Desert. The main goal of the proposed system is to support the best utilization of farm resources. In this research, predicting the best sowing dates for the aimed crops is based on weather conditions prediction along with calculating the seasonal accumulative growing degree days (GDD) fulfillment duration for each crop. Various Machine Learning (ML) regression algorithms have been used for predicting the daily minimum and maximum air temperature based on historical weather conditions data for twenty-five growing seasons (1990/91 to 2014/15). Experimental results showed that using the M5P and IBk ML regression algorithms have outperformed the other implemented regression algorithms for predicting the daily minimum and maximum air temperature based on historical weather conditions data. That has been measured based on the calculated mean absolute error (MAE). Also, obtained experimental results obviously indicated that the best cultivation-time prediction by the proposed recommender system has been achieved by the M5P algorithm, based on the seasonal accumulative GDD fulfillment duration, for the coming five growing seasons (2016/17 to 2019/20).