Machine Learning Enabled Image Classification Using K-Nearest Neighbour and Learning Vector Quantization

Nowadays, the classification of images is used to bridge the gap between human vision as well as computer vision to identify images by machines in the same way humans do. It concerns assigning the appropriate class for a provided image. The major problems encountered in the classification of images is the representation of image vector as well as image feature extraction. To overcome major image classification issues, machine learning algorithms such as K-NN and LVQ are implemented using Scikit-learn and Python packages on the Iris dataset. LVQ and KNN models were developed using a 70:30 split ratio for training and testing the models. The algorithm's performance was compared using machine learning performance metrics such as Accuracy, F1 Score, Recall, and Precision. According to the findings, KNN is a superior option for classification tasks that demand high accuracy-based parameter setting, and the implementation followed specifically due to the metrics result with 96.67% accuracy, 1.00 precision, 0.89 recall, and 0.94 F1 Score. Hence, KNN can be applied for effective image classification problems. KNN and LVQ both have their strength and weaknesses depending on the problem at hand. The study, therefore, recommends the use of other machine learning algorithms such as random forest, and decision tree with hybridization of ensemble learning on the same dataset for optimal comparison.