AI Paper Atlas

Research map / Label-efficient and robust learning / Domain adaptation and generalization

Cross-Domain Semantic Segmentation on Inconsistent Taxonomy using VLMs

Jeongkee Lim, Yusung Kim

ECCV 2024 · Label-efficient and robust learning › Domain adaptation and generalization · approach: Adaptation for semantic segmentation · 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
add CSI (VLM-based relabeling with OWL-ViT and CLIP) to MICMIC · Synthia -> Cityscapes, 19 classesmIoU19: 56.7 -> 67.4Improved
add CSI to MICMIC · Synthia -> Cityscapes, 16 shared classesmIoU16: 67.3 -> 68.8Improved
add CSI to DAFormerDAFormer · Synthia -> CityscapesmIoU19: 51.3 -> 60.0Improved
add CSI to CNN-based DACSDACS (DeepLabV2) · Synthia -> CityscapesmIoU19: 40.7 -> 45.5Improved
use OWL-ViT detection without CLIP validation (ablation)CSI · Synthia -> CityscapesmIoU16: 67.3 -> 64.8Hurt
add CLIP validation to OWL-ViT detection (ablation)CSI · Synthia -> CityscapesmIoU19: 60.6 -> 67.4Improved
start relabeling too late in training (ablation)CSI · Synthia -> CityscapesmIoUHurt
start relabeling too early in training (ablation)CSI · Synthia -> CityscapesmIoUHurt
add CSI to MIC on harsher target domainMIC · Synthia -> ACDCmIoU16: 44.7 -> 46.2Improved
use source with more similar classes (fully labeled GTA) versus CSI on label-removed GTAMIC · GTA -> CityscapesmIoU19: 73.1 (MIC*+CSI) vs 75.9 (MIC)Mixed

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 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