Fashion Image Classification Using Convolutional Neural Network-VGG16 and eXtreme Gradient Boosting Classifier

Deep learning models have been gaining importance in recent years and are mostly used in various computer vision applications. Because of its ability to extract complex features from input images, deep learning constitutes an efficient tool for performing image recognition and classification. Apparel classification is considered important field of research of computer vision that has been explored and investigated. This work proposes an architecture based on CNN-VGG16 as a trainable feature extractor and XGBoost as a classifier. The input image is preprocessed and then extracted by CNN-VGG16 to produce features that XGBoost uses to produce a result. The proposed model was evaluated on the fashion MNIST dataset. The results obtained show that our proposed method performs better than other methods described in the literature with the same dataset.