Coloring With Limited Data: Few-Shot Colorization via Memory-Augmented Networks
This addresses the data efficiency issue in automatic colorization for applications like image editing, though it is incremental as it builds on existing memory network ideas.
The paper tackled the problem of few-shot colorization in deep learning by introducing MemoPainter, a memory-augmented model that produces high-quality colorization with limited data, achieving superior quality in few-shot and one-shot tasks.
Despite recent advancements in deep learning-based automatic colorization, they are still limited when it comes to few-shot learning. Existing models require a significant amount of training data. To tackle this issue, we present a novel memory-augmented colorization model MemoPainter that can produce high-quality colorization with limited data. In particular, our model is able to capture rare instances and successfully colorize them. We also propose a novel threshold triplet loss that enables unsupervised training of memory networks without the need of class labels. Experiments show that our model has superior quality in both few-shot and one-shot colorization tasks.