Abstract:Automatic colorization of gray images with objects of different colors and sizes is challenging due to inter- and intra-object color variation and the small area of the main objects due to extensive backgrounds. The learning process often favors dominant features, resulting in a biased model. In this paper, we formulate the colorization problem into a multinomial classification problem and then apply a weighted function to classes. We propose a set of formulas to transform color values into color classes and vice versa. Class optimization and balancing feature distribution are the keys for good performance. Observing class appearance on various extremely large-scale real-time images in practice, we propose 215 color classes for our colorization task. During training, we propose a class-weighted function based on true class appearance in each batch to ensure proper color saturation of individual objects. We establish a trade-off between major and minor classes to provide orthodox class prediction by eliminating major classes' dominance over minor classes. As we apply regularization to enhance the stability of the minor class, occasional minor noise may appear at the object's edges. We propose a novel object-selective color harmonization method empowered by the SAM to refine and enhance these edges. We propose a new color image evaluation metric, the Chromatic Number Ratio (CNR), to quantify the richness of color components. We compare our proposed model with state-of-the-art models using five different datasets: ADE, Celeba, COCO, Oxford 102 Flower, and ImageNet, in both qualitative and quantitative approaches. The experimental results show that our proposed model outstrips other models in visualization and CNR measurement criteria while maintaining satisfactory performance in regression (MSE, PSNR), similarity (SSIM, LPIPS, UIQI), and generative criteria (FID).

COCO-LC: Colorfulness Controllable Language-based Colorization

Language-based colorization of scene sketches

L-CoDe: Language-Based Colorization Using Color-Object Decoupled Conditions

L-C4: Language-Based Video Colorization for Creative and Consistent Color

L-CoDer: Language-Based Colorization with Color-Object Decoupling Transformer

Control Color: Multimodal Diffusion-based Interactive Image Colorization

L-CAD: Language-based Colorization with Any-level Descriptions Using Diffusion Priors

L-CoIns: Language-Based Colorization with Instance Awareness

Improving reference-based image colorization for line arts via feature aggregation and contrastive learning

LUCSS: Language-based User-customized Colourization of Scene Sketches

Self-driven Dual-path Learning for Reference-based Line Art Colorization under Limited Data

Two-stage image colorization via color codebook

DiffColor: Toward High Fidelity Text-Guided Image Colorization with Diffusion Models

SCSNet: an Efficient Paradigm for Learning Simultaneously Image Colorization and Super-resolution

Improved Diffusion-based Image Colorization via Piggybacked Models

Language-based Photo Color Adjustment for Graphic Designs

Image Colorization using CycleGAN with semantic and spatial rationality

Natural Grayscale Image Colorization via Local Sparse Coding

Learning to Color from Language

CCC: Color Classified Colorization

Towards Vivid and Diverse Image Colorization with Generative Color Prior