Fine-Tuning Image-Conditional Diffusion Models is Easier than You Think discuss: huggingface.co/papers/2409.11… Recent work showed that large diffusion models can be reused as highly precise monocular depth estimators by casting depth estimation as an image-conditional image

Hacker News of AI research papers built with o1-preview/mini in Gradio, added instructions on how to submit a paper

Great work by my team at RWTH Computer Vision Group. Up to now, the prevailing perception was that diffusion models must be slow or complex for depth estimation. We show that this is not the case. With our training scheme, faster more efficient diffusion-based models are possible.

Fine-Tuning Image-Conditional Diffusion Models is Easier than You Think Gonzalo Martin Garcia Karim Abou Zeid Christian Schmidt Daan de Geus Alexander Hermans Bastian Leibe tl;dr: there is a bug in Marigold noise DDIM scheduler, if you fix ot, you can do 1 step depth estimation. arxiv.org/abs/2409.11355

This is the most exciting paper I've read in a while. Alternate title could have been: "One weird trick to increase your depth map inference 200x." arxiv: arxiv.org/abs/2409.11355 github: github.com/VisualComputin… Let's go through the details 🧵 1/9

Check out our recent work on interactive 4D #LiDAR sequence segmentation. Our #interactive4D approach improves upon previous methods, resulting in higher segmentation quality with fewer clicks, while providing consistent object instance IDs across time.