Spaces:
Running
on
Zero
Apply for community grant: Academic project (gpu)
Repurposing Diffusion-Based Image Generators for Monocular Depth Estimation
This demo space complements our paper titled "Repurposing Diffusion-Based Image Generators for Monocular Depth Estimation" by Bingxin Ke, Anton Obukhov, Shengyu Huang, Nando Metzger, Rodrigo Caye Daudt, and Konrad Schindler.
It allows the user to estimate an affine-invariant depth map from an input RGB image. Our method is based on the Stable Diffusion pipeline and is notable for its drastically improved performance on in-the-wild images over established baselines, such as MiDaS and LeRes. This is one more reason we would like to showcase our method in HF Spaces and engage with the community.
Processing user uploads requires an A10G small GPU. Persistence is disabled. The demo is powered by Gradio v3.
Hi @toshas , we have assigned a gpu to this space. Note that GPU Grants are provided temporarily and might be removed after some time if the usage is very low.
To learn more about GPUs in Spaces, please check out https://huggingface.co/docs/hub/spaces-gpus