Spaces:
Running
on
Zero
Apply for community grant: Academic project (gpu and storage)
Project: GeoRemover — Removing Objects and Their Causal Visual Artifacts
Paper: https://arxiv.org/abs/2509.18538
Summary: GeoRemover is a two-stage diffusion framework for realistic object removal.
Stage-1 (Geometry Removal): structure-aware inpainting guided by depth/geometry priors to keep scene layout stable.
Stage-2 (Appearance Rendering): appearance synthesis that respects illumination, textures, shadows/reflections, and other causal effects.
Tech stack: Hugging Face Diffusers (FLUX Fill) with a modified forward and LoRA; packaged as a reproducible pipeline with configurable parameters.
Demo goals: enable hands-on evaluation, teaching, and fair benchmarking via a simple Gradio UI (image + mask → result, with side-by-side visualization and downloads).
Why GPU: high-resolution inference benefits from an A100-class GPU to keep latency practical; a ZeroGPU (free A100) grant is suitable for day-to-day academic usage.
Artifacts: model repo on the Hub; the Space links model cards, tags, and (upon approval) the paper page.
Hi
@buxiangzhiren
, we've assigned ZeroGPU to this Space. Please check the compatibility and usage sections of this page so your Space can run on ZeroGPU.
If you can, we ask that you upgrade to Pro ($9/month) to enjoy higher ZeroGPU quota and other features like Dev Mode, Private Storage, and more: hf.co/pro