AI Paper Atlas

Research map / 3D vision / Depth and stereo

Self-Supervised Relative Depth Learning for Urban Scene Understanding

Huaizu Jiang, Gustav Larsson, Michael Maire Greg Shakhnarovich, Erik Learned-Miller

ECCV 2018 · 3D vision › Depth and stereo · approach: Dense depth prediction and completion · 83 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
add relative depth self-supervised pre-trainingAlexNet FCN32s · CityScapes segmentationmean IoU: 40.7 -> 45.4Improved
add relative depth self-supervised pre-trainingAlexNet FCN32s · CamVid segmentationmean IoU: 44.0 -> 53.4Improved
add relative depth pre-training on CityDrivingVGG16 FCN8s · CityScapes segmentationmean IoU: 51.6 -> 59.8Improved
add relative depth pre-training on CityDrivingVGG16 FCN8s · KITTI segmentationmean IoU: 34.2 -> 48.6Improved
replace ImageNet supervised pre-training with relative depth pre-trainingVGG16 FCN8s · CityScapes segmentationmean IoU: 63.4 -> 59.8Hurt
add unsupervised domain adaptation by fine-tuning depth task on CityScapes videos (ablation)Ours CD · CityScapes FCN32smean IoU: 55.0 -> 56.2Improved
add unsupervised domain adaptation by fine-tuning depth task on KITTI videos (ablation)Ours CD · KITTI FCN32smean IoU: 45.6 -> 46.0Improved
add unsupervised domain adaptation on KITTI videos (ablation)Ours CD · CamVid FCN8smean IoU: 60.3 -> 60.1No clear effect
add domain adaptation on KITTI videos (ablation)Ours CD · road segmentation F1F1: 94.74 -> 95.66Improved
add domain adaptation on KITTI videos (ablation)Ours CD · car detection easyAP: 92.84 -> 94.31Improved
replace ImageNet pre-training with relative depth pre-training plus KITTI adaptationVGG16 U-Net · KITTI monocular depthRMSE: 5.546 -> 5.317Improved
replace ImageNet pre-training with relative depth pre-training plus KITTI adaptationResNet50 U-Net · KITTI monocular depthRMSE: 5.073 -> 4.903Improved

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