AI Paper Atlas

Research map / Vision-language and multimodal / Cross-modal learning

Beyond What's Shared: Recovering Lost Unique Information from Intermediate Layers to Boost Multimodal Geo-Foundation Models

JangHyeon Lee, Philipe Ambrozio Dias, Yao-Yi Chiang, Dalton Lunga

CVPR 2026 · Vision-language and multimodal › Cross-modal learning · approach: Multimodal fusion with missing modalities

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
fuse intermediate and final layer embeddings (BWS)SatCLIP · Biometop-1 accuracy: 0.848 -> 0.911Improved
fuse intermediate and final layer embeddings (BWS)CSP · ElevationR2: 0.390 -> 0.659Improved
fuse intermediate and final layer embeddings (BWS)GeoCLIP · ElevationR2: 0.608 -> 0.809Improved
add BWS on top of RANGERANGE · Ecoregionstop-1 accuracy: 0.894 -> 0.920Improved
add BWS on top of RANGERANGE · TemperatureR2: 0.894 -> 0.897No clear effect
add BWS on top of RANGERANGE · HousingR2: 0.449 -> 0.457No clear effect
use BWS layer fusion instead of dimension-matched concatenation with GeoCLIP (ablation)SatCLIP · Seven-task averageaverage performance: 0.748 -> 0.788Improved
use BWS layer fusion instead of random projection of same dimension (ablation)SatCLIP · Seven-task averageaverage performance: 0.750 -> 0.788Improved
apply BWS to a contrastively trained (multimodal) image encoder rather than unimodal (ablation)MoCo-ResNet50 image encoder · Seven-task averageaverage performance: 0.247 -> 0.301 unimodal vs 0.523 -> 0.700 multimodalImproved
use uniform layer weights (alpha=0) instead of emphasizing early or late layers (ablation)BWS · Seven-task averageoverall average: 0.788 vs 0.781 (alpha -2.0) and 0.776 (alpha 2.0)Improved
use middle layers instead of late layers only (ablation)SatCLIP · Seven-task averageoverall average: 0.755 -> 0.780Improved
increase encoder depth when using BWS (ablation)SatCLIP with BWS · Seven-task averageoverall average: 0.790 -> 0.826Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

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