BS723 | Machine Learning Applications in Architecture
This project focuses on the comparative analysis of image classification using coloured and greyscale datasets. The main objective is to explore the impact of colour information on the performance of image classification models. The project consists of three distinct phases: Phase I, Phase II, and Phase III, each involving specific modifications to the dataset and training processes.
In Phase I, a deep learning model based on the VGG16 architecture was trained on a dataset of colored images. The dataset was divided into training and validation sets, and the model’s performance was evaluated using various metrics such as accuracy, precision, recall, and F1-score. This phase served as the baseline for further analysis.
In Phase II, the test data was converted to greyscale while keeping the model architecture and training process the same. This allowed for the examination of the influence of colour on the model’s performance by comparing the results with Phase I.
Phase III involved converting the entire dataset to greyscale, including both the training and testing data. This allowed for a comprehensive understanding of the model’s performance when trained and evaluated solely on greyscale images.