Partial Code Implementation for Artistic Style and Emotion Prediction Using CNN and Vision Transformers

This repository provides a partial implementation of the methods discussed in our paper, "Joint Prediction of Artistic Image Style Categories and Audience Emotions Based on CNN, ViT, and Multi-Task Learning." Due to confidentiality constraints, this repository includes only a subset of the full codebase. The available code illustrates the core principles and methodologies used to achieve joint predictions of artistic style and audience emotions. It includes preprocessing utilities, model configuration examples, and basic training routines.