Reproduce “EagleDepth: Efficient Fine-Grained Depth Estimation via Pixel Diffusion Decoder” 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
Bowen Chai, Tianbao Zhang, Shuyu Wu, Dexin Zuo, Zhaoxin Fan, Danping Zou
EagleDepth estimates high-resolution monocular depth by pairing a low-resolution latent model for scene-wide geometry with a pixel-space decoder that uses the original RGB image for detail. The latent model runs at 512² pixels, and the decoder generates target-resolution depth without its VAE decoder at inference. On synthetic Synth4K, it leads high-frequency and boundary metrics; real-world tests use approximately 1024² inputs.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.