AI Paper Atlas

Research map / 3D vision / Depth and stereo

Visualization of Convolutional Neural Networks for Monocular Depth Estimation

Junjie Hu, Yan Zhang, Takayuki Okatani

ICCV 2019 · 3D vision › Depth and stereo · approach: Dense depth prediction and completion · 94 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
increase sparseness weight lambda from 1 to 6 (ablation)mask predictor G for depth CNN · NYU-v2 (ResNet-50 target, mask-sparsity sweep)RMSE (continuous mask M): 0.605 -> 0.772Hurt
increase sparseness weight lambda from 1 to 6 (ablation)mask predictor G for depth CNN · NYU-v2 (ResNet-50 target, mask-sparsity sweep)Sparseness (fraction non-zero pixels): 0.920 -> 0.215Improved
increase lambda from 1 to 6, binarized mask (ablation)mask predictor G for depth CNN · NYU-v2 (ResNet-50 target, mask-sparsity sweep)RMSE (binarized mask M'): 0.568 -> 0.882Hurt
binarize mask at small lambda (lambda=1) (ablation)mask predictor G for depth CNN · NYU-v2 (ResNet-50 target, mask-sparsity sweep)RMSE: 0.605 -> 0.568Improved
binarize mask at large lambda (lambda=6) (ablation)mask predictor G for depth CNN · NYU-v2 (ResNet-50 target, mask-sparsity sweep)RMSE: 0.772 -> 0.882Hurt
binarize mask at lambda=5 (ablation)mask predictor G for depth CNN · NYU-v2 (ResNet-50 target, mask-sparsity sweep)RMSE: 0.740 -> 0.758Hurt
mask only 36.1% of pixels (lambda=5) vs full image (ablation)mask predictor G for depth CNN · NYU-v2 (ResNet-50 target, mask-sparsity sweep)RMSE: 0.555 -> 0.740Hurt

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