Reproduce “Fast End-to-End Trainable Guided Filter” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Huikai Wu, Shuai Zheng, Junge Zhang, Kaiqi Huang
DGF, or Deep Guided Filter, speeds dense pixel-wise prediction by running a fully convolutional network on a downsampled image, then refining its output at full resolution with learned, edge-aware guidance from the original image. On MIT–Adobe FiveK style transfer, the full DGF model reaches 26.17 dB PSNR; at 2048², it takes 21 ms per image versus CAN’s 160 ms on the reported setup.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.