AI Paper Atlas

Research map / Vision-language and multimodal / Multimodal LLMs

VisualEchoes: Spatial Image Representation Learning through Echolocation

Ruohan Gao, Changan Chen, Ziad Al-Halah, Carl Schissler, Kristen Grauman

ECCV 2020 · Vision-language and multimodal › Multimodal LLMs · approach: Spatial reasoning in VLMs · 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 echo input to RGB depth predictionRGB2Depth · Replica depth prediction (heldout, case study)RMS: 0.374 -> 0.346Improved
use echoes only instead of average depthEcho2Depth · Replica depth prediction (heldout, case study)RMS: 1.070 -> 0.713Improved
pre-train encoder with VisualEchoes instead of training from scratchRGB2Depth UNet · Replica depthRMS: 0.360 -> 0.332Improved
pre-train encoder with VisualEchoes instead of training from scratchRGB2Depth UNet · NYU-V2 depth (UNet)RMS: 0.818 -> 0.797Improved
pre-train encoder with VisualEchoes instead of training from scratchRGB2Depth UNet · DIODE depthRMS: 2.352 -> 2.223Improved
pre-train with ImageNet classification instead of VisualEchoesRGB2Depth UNet · Replica depthRMS: 0.332 -> 0.356Hurt
use 4-class orientation prediction instead of 2-class (ablation)VisualEchoes · Replica depthRMS: 0.340 -> 0.332Improved
use orientation prediction instead of binary echo-image matching (ablation)VisualEchoes · Replica depthRMS: 0.345 -> 0.332Improved
pre-train ResNet-50 with VisualEchoes instead of scratchdepth network · NYU-V2 depthRMS: 0.804 -> 0.683Improved
pre-train with ImageNet instead of VisualEchoesdepth network · NYU-V2 depthRMS: 0.683 -> 0.555Hurt
pre-train PSPNet with VisualEchoes instead of scratchPSPNet · NYU-V2 surface normalsmean angle distance: 26.3 -> 22.9Improved
pre-train navigation CNN with VisualEchoes instead of scratchPointGoal agent · Replica navigationSPL: 0.830 -> 0.856Improved

Later papers that compared with this method

1 later accepted paper reports VisualEchoes as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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