AI Paper Atlas

Research map / Architectures and efficiency / Pruning and efficient tuning

Parameter-Efficient Adaptation of Geospatial Foundation Models through Embedding Deflection

Romain Thoreau, Valerio Marsocci, Dawa Derksen

ICCV 2025 · Architectures and efficiency › Pruning and efficient tuning · approach: LoRA variants and theory · 8 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
adapt only the UPerNet layers instead of all layers with uAtt (ablation)DEFLECT · MADOS, Scale-MAEmIoU: 38.7 -> 50.6Improved
adapt UPerNet layers instead of only the first layer (ablation)DEFLECT · MADOS, Scale-MAEmIoU: 47.1 -> 50.6Improved
adapt UPerNet layers instead of only the first layer (ablation)DEFLECT · MADOS, Cross-Scale MAEmIoU: 39.2 -> 38.2Hurt
remove low-rank reparameterization of uAtt parameters (ablation)DEFLECT · MADOS, DINO-MCmIoU: 51.6 -> 52.9Improved
increase LoRA rank from 16 to 32 (competitor) (ablation)LoRA · MADOS, Scale-MAEmIoU: 45.0 -> 30.2Hurt
add linear projection layer to spectral embeddings (ablation)DEFLECT · MADOS, DINO-MCmIoU: 48.5 -> 51.6Improved
sample 50% of pixels instead of 10% in pixel-set encoding (ablation)DEFLECT · MADOS / BurnScarsmIoU / IoU: 50.6 -> 51.8 (Scale-MAE MADOS)No clear effect
replace LoRA/SLR with DEFLECT on DINO-MCLoRA / SLR · Average performance (5 datasets, DINO-MC)Avg. performance: 52.7 -> 64.8 (vs SLR)Improved
use DEFLECT instead of full finetuningCross-Scale MAE · ForestNet (mF1)mF1: 37.1 -> 46.3Improved
use DEFLECT instead of full finetuningScale-MAE · MADOS (mIoU)mIoU: 47.0 -> 50.6Improved
use RGB+random initialization instead of Repeat for new bands (ablation)PEFT methods · BurnScars / MADOSIoU / mIoU: 70.5 -> 73.2 (SLR, DINO-MC, BurnScars)Mixed

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