The aim of the project is to classify images of Nike, Adidas and Converse shoes using CNNs. Due to small size of the dataset transfer learning was used to create a classification model.
The dataset was obtained by downloading images from Google Images. It has 2 folders: one with the train data and one with the test data. The train-test split ratio is 0.14, with the train dataset containing 711 images and the test dataset containing 114 images. All images are 240x240 pixels in RGB color model.
As a result of the project a convolutional neural network was built.
The accuracy of the model is above 0.98 on the training set and 0.9211 on the test set.