What is it about?
Humans can quickly learn handwriting imitation with hallucinations, while this task is challenging for machines. Humans can imitate handwriting perhaps because they can learn to disentangle textual contents and calligraphic styles. Inspired by this, we propose HiGAN+ for realistic handwritten text synthesis based on disentangled representations, achieving state-of-the-art performance.
Featured Image
Photo by Dan Cristian Pădureț on Unsplash
Why is it important?
Although machines can easily recognise humans’ handwriting scripts with recent advanced techniques, it still remains challenging for machines to synthesise realistic handwriting images. Hence, it will step closer to high-level artificial intelligence if we can teach machines/robotics to write texts as realistic as humans.
Perspectives
Humans’ handwriting is very arbitrary and thus HiGAN+ indeed has limits for synthesizing meaningful handwriting images. Nevertheless, it is interesting to teach machines/robotics to write texts as realistic as humans, which takes a closer step to high-level artificial intelligence.
Ji Gan
Chongqing University of Posts and Telecommunications
Read the Original
This page is a summary of: HiGAN+: Handwriting Imitation GAN with Disentangled Representations, ACM Transactions on Graphics, February 2023, ACM (Association for Computing Machinery),
DOI: 10.1145/3550070.
You can read the full text:
Contributors
The following have contributed to this page
