A Study on Color Theme Generation Using Convolutional Neural Networks

This study explores an innovative method for generating color themes from images using Convolutional Neural Networks (CNNs). Grounded in the integration of the Munsell color system for accurate color categorization and Mini Batch K-means for effective color quantization, the study advances the automation and enhancement of color theme generation. The novel approach emphasizes training separate ResNet models for different colors, subsequently merging them to capture a broad and aesthetically pleasing color spectrum. Results indicate a significant improvement in the accuracy and diversity of the automatically generated color themes, surpassing traditional manual and semi-automated methods. This method not only increases the efficiency of color theme generation but also introduces a scalable model adaptable to various applications in digital media and design industries.