AI Paper Atlas

Research map / Vision-language and multimodal / Human and embodied modeling

DIP: Unsupervised Dense In-Context Post-training of Visual Representations

Sophia Sirko-Galouchenko, Spyros Gidaris, Antonin Vobecky, Andrei Bursuc, Nicolas Thome

ICCV 2025 · Vision-language and multimodal › Human and embodied modeling · approach: Vision-language pretraining · 0 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
post-train DINOv2R with DIPDINOv2R ViT-B/14 · PascalVOC (1/128 of support data)mIoU: 60.3 -> 70.1Improved
post-train DINOv2R with DIPDINOv2R ViT-B/14 · ADE20K (in-context, full data)mIoU: 40.8 -> 42.6Improved
apply DIP post-training to MAEMAE ViT-B/16 · PascalVOC (in-context)mIoU: 13.9 -> 47.3Improved
apply DIP post-training to CLIPCLIP ViT-B/16 · PascalVOC (in-context)mIoU: 71.8 -> 73.8Improved
apply DIP post-training to DINOv2R ViT-LDINOv2R ViT-L · PascalVOC (full data)mIoU: 76.9 -> 81.1Improved
use in-context objective instead of direct dense prediction (ablation)DIP · NYUv2 depthRMSE: .776 -> .756Improved
use in-context objective instead of direct dense prediction (ablation)DIP · ADE20KmIoU: 39.9 -> 39.7No clear effect
add distractor support examples (ablation)DIP · PascalVOCmIoU: 78.5 -> 81.0Improved
use DINOv2R nearest-neighbor positives instead of two random crops (ablation)DIP · PascalVOCmIoU: 79.5 -> 81.0Improved
add DiffCut segments before K-means pseudo-labeling (ablation)DIP · PascalVOCmIoU: 59.1 -> 81.0Improved
use ground truth labels instead of pseudo-labels (ablation)DIP · PascalVOCmIoU: 81.0 -> 81.9No clear effect
vary number of pseudo-classes (300 to 2000) (ablation)DIP · PascalVOCmIoU: 80.8 -> 81.0No clear effect

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