Recognition and classification of mechanical tools through machine learning

Abstract

This thesis presents a study on recognizing mechanical tools by classification through machine learning. The research is to develop a model that can accurately classify several mechanical tools based on their physical characteristics. The study starts with a literature review of earlier work on machine learning for object recognition and classification and a survey of existing mechanical tool classification systems. The Kaggle dataset of mechanical tools with their labelling will be used in this research. The dataset is used to train and evaluate three different machine learning models i.e., VGG16, LeNet, and AlexNet, and compare the results of these models regarding classification accuracy.

The study also explores the impact of different pre-processing and feature extraction techniques on classification performance. The thesis concludes with a discussion of the dataset used, the method applied, the results obtained, the implications of the research, and potential future work in this area