VCGAN: Video Colorization with Hybrid Generative Adversarial Network

TL;DR

VCGAN is a novel end-to-end hybrid GAN model that improves video colorization by enhancing temporal consistency and integrating colorization and refinement into a single architecture, leading to higher quality results.

Contribution

The paper introduces VCGAN, a hybrid recurrent GAN that unifies colorization and refinement networks, and incorporates global and placeholder feature extractors for better quality and temporal consistency.

Findings

VCGAN outperforms existing methods in color quality.

VCGAN achieves superior temporal consistency in videos.

The dense long-term loss improves far-frame coherence.

Abstract

We propose a hybrid recurrent Video Colorization with Hybrid Generative Adversarial Network (VCGAN), an improved approach to video colorization using end-to-end learning. The VCGAN addresses two prevalent issues in the video colorization domain: Temporal consistency and unification of colorization network and refinement network into a single architecture. To enhance colorization quality and spatiotemporal consistency, the mainstream of generator in VCGAN is assisted by two additional networks, i.e., global feature extractor and placeholder feature extractor, respectively. The global feature extractor encodes the global semantics of grayscale input to enhance colorization quality, whereas the placeholder feature extractor acts as a feedback connection to encode the semantics of the previous colorized frame in order to maintain spatiotemporal consistency. If changing the input for placeholder feature extractor as grayscale input, the hybrid VCGAN also has the potential to perform image colorization. To improve the consistency of far frames, we propose a dense long-term loss that smooths the temporal disparity of every two remote frames. Trained with colorization and temporal losses jointly, VCGAN strikes a good balance between color vividness and video continuity. Experimental results demonstrate that VCGAN produces higher-quality and temporally more consistent colorful videos than existing approaches.