Classification of Flower Species based on Flower Texture to Improve Accuracy of Classifier using Linear Regression and Comparing with SVM algorithm.

Abstract

Aim: Classification of flower species based on innovative flower texture to improve accuracy of classifier using linear regression and comparing with SVM algorithm. Methods and Materials: Flower species recognition is performed using Linear Regression (N=10) over SVM (N=10) with the split size of training and testing dataset 70% and 30% respectively. Calculation of samples is done by using G power of 80% which contains two different groups, alpha (0.05), power (80%) and environment ratio 1. Results: Linear Regression has significantly better accuracy (95.9%) compared to SVM (93.3%) and attained significance value of p = 0.01. Conclusion: Linear regression achieved significantly better flower recognition than SVM for identifying the different types of flower species.