AI Paper Atlas

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

MapFormer: Boosting Change Detection by Using Pre-change Information

Maximilian Bernhard, Niklas Strauß, Matthias Schubert

ICCV 2023 · Vision-language and multimodal › Cross-modal learning · approach: Multi-modality image fusion · 18 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 pre-change semantic map as input (concatenation baseline)Concatenation change detector · DynamicEarthNetbinary change IoU: 11.6 -> 20.1Improved
add pre-change semantic map as input (concatenation baseline)Concatenation change detector · HRSCDbinary change IoU: 25.9 -> 45.1Improved
use MapFormer fusion module instead of plain concatenationConditional BCD · DynamicEarthNetbinary change IoU: 20.1 -> 23.5Improved
use MapFormer fusion instead of DynamicMLPConditional BCD · HRSCDbinary change IoU: 45.7 -> 47.7Improved
add supervised cross-modal contrastive loss (ablation)MapFormer K=10 · DynamicEarthNetbinary change IoU: 20.9 -> 23.5Improved
add stop-gradient on map features in contrastive loss (ablation)MapFormer K=10 · DynamicEarthNetbinary change IoU: 22.5 -> 23.5Improved
do not add linear projection on map features (ablation)MapFormer K=10 · HRSCDbinary change IoU: 45.9 -> 47.7Improved
use full-class semantic map instead of merged high-level classes (ablation)MapFormer K=10 · DynamicEarthNetbinary change IoU: 17.6 -> 23.5Improved
use full-resolution semantic map instead of 32x downsampled (ablation)MapFormer K=10 · HRSCDbinary change IoU: 43.2 -> 47.7Improved
vary hyperparameter K (5, 10, 15) (ablation)MapFormer · HRSCDbinary change IoU: 48.0 / 47.7 / 47.6No clear effect
use predicted maps instead of true pre-change map (w/o m1)MapFormer K=10 · DynamicEarthNet (bi-temporal)binary change IoU: 11.8 -> 16.7Improved
drop pre-change image (cross-modal) and keep pre-change map (ablation)MapFormer · DynamicEarthNetbinary change IoU: 23.5 -> 19.0Hurt

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