AI Paper Atlas

Research map / 3D vision / Depth and stereo

Diffusion Models for Monocular Depth Estimation: Overcoming Challenging Conditions

Fabio Tosi, Pierluigi Zama Ramirez, Matteo Poggi

ECCV 2024 · 3D vision › Depth and stereo · approach: Self-supervised monocular depth · 34 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
fine-tune monocular networks on diffusion-generated night and rain images with depth pseudo-labelsDPT · nuScenes nightabsRel: 0.354 -> 0.224Improved
fine-tune Depth Anything with diffusion-distilled dataDepth Anything · nuScenes nightabsRel: 0.291 -> 0.219Improved
fine-tune ZoeDepth with diffusion-distilled dataZoeDepth · nuScenes day-clearRMSE: 8.517 -> 8.946Hurt
fine-tune md4all with diffusion-generated data instead of GAN day-to-night translation requiring target imagesmd4all-DD · nuScenes nightabsRel: 0.242 -> 0.188Improved
fine-tune md4all-DD with diffusion data on RobotCar nightmd4all-DD · RobotCar nightabsRel: 0.391 -> 0.139Improved
use Depth Anything or DPT instead of md4all depth to condition the diffusion model (ablation)md4all-DD ft. Ours · nuScenes day-rainabsRel: 0.147 -> 0.139 (Depth Anything); 0.143 (DPT)Improved
fine-tune on generated adverse images from only daytime datasets (Mapillary, Cityscapes, KITTI, Apolloscape)DPT · DrivingStereo day-rainabsRel: 0.188 -> 0.124Improved
fine-tune on daytime-only generated dataDepth Anything · DrivingStereo day-rainabsRel: 0.112 -> 0.110Improved
fine-tune on text-generated transparent and mirror objects instead of real ToM data and GT masksDPT-Large · Booster ToM pixelsdelta<1.25 (%): 96.17 vs 96.68 for Costanzino et al.No clear effect
fine-tune Depth Anything on generated non-Lambertian objectsDepth Anything · Booster ToM pixelsdelta<1.25 (%): 84.23 -> 98.91Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy