A Systems Approach to Validating and Analyzing Improvements to Real-Time Image Classification Utilizing Machine Learning

摘要

This paper delves into the generation and use of image classification models in a real-time environment utilizing machine learning. The ImageAI framework is used to generate a list of models from a set of training images and also for classifying new images using the generated models. Through this paper, previous research projects and industry programs are analyzed for design and operation. The basic implementation results in models that classify new images correctly the majority of the time with a high level of confidence. However, almost a quarter of the time the models classify images incorrectly. This paper attempts to improve the classification accuracy and improve the operational efficiency of the overall system as well.