Классификация изображений для извлечения данных о моде человека

With the always increasing amount of image data, it has become a necessity to automatically look for and process information in these images. As fashion is captured in images, the fashion sector provides the perfect foundation for image classification. In this thesis, the state-of-the art for image classification is analyzed and discussed. Based on the elaborated knowledge, four different approaches will be implemented to successfully extract features out of fashion data. For this purpose, a human worn fashion dataset with 2567 images was created. The results show, that convolutional neural networks are the undisputed standard for classifying images, and that TensorFlow is the best library to build them. The results show, that through the introduction of dropout layers, data augmentation and transfer learning, the model was successfully prevented from overfitting and it was possible to incrementally improve the validation accuracy on the created dataset from an initial 69% to a final validation accuracy of 84%. More distinct apparels like trousers, shoes and hats were better classified than other upper body clothes.