Fashion Art Style Analysis Using Inception V3 and Feature Extraction

Summary: This project utilizes a lightweight and computationally efficient Inception V3 model as the baseline model. It crawls the internet to collect high-discussed fashion art style images and conducts standardized training. Feature vectors are extracted from the generated model and scatter plots are drawn. The project analyzes the scatter plots to elucidate the correlations and differences between different fashion styles by representative feature vectors. It also conducts inference analysis on the preferences of specific fashion style characteristics for target audiences, aiming to provide effective information recommendations and expand consumer base.

Keywords: fashion art style, Inception V3, convolutional neural network, feature extraction, information recommendation.