AI Paper Atlas

Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning

Zero-Shot Image Feature Consensus with Deep Functional Maps

Xinle Cheng, Congyue Deng, Adam Harley, Yixin Zhu, Leonidas Guibas

ECCV 2024 · Label-efficient and robust learning › Zero-/few-shot and meta-learning · approach: Transductive and semantic-dictionary zero-shot · 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 diagonal compactness and consistency regularizers to descriptor loss (ablation)Feature functional map · SPair-71kPCK@0.1: 44.6 -> 55.3Improved
add both regularizers on top of one regularizer (ablation)Feature functional map · SPair-71kMSE: 97.9 -> 88.0Improved
add only a single regularizer (diag or cons) over no regularization (ablation)Feature functional map · SPair-71kMSE: 95.3 -> 97.9 / 100.3Hurt
use different feature sets for basis (SD) and loss (DINOv2) instead of the same set (ablation)Feature functional map · TSSPCK avg: 71.0 -> 78.9Improved
choose SD for basis and DINOv2 for loss rather than reverse (ablation)Feature functional map · TSSPCK avg: 72.2 -> 78.9Improved
use functional map with a single feature set versus raw nearest neighbors (ablation)DINOv2-ViT-B/14 · TSSPCK avg: 69.4 -> 71.0Improved
fuse features via functional map instead of concatenationDINOv2-ViT-B/14 + SD · TSSsmoothness: 6.81 -> 5.74Improved
fuse DINOv2-ViT-S/14 and SD via functional map instead of concatenationDINOv2-ViT-S/14 + SD · TSSPCK@0.05: 78.1 -> 71.5Hurt
fuse different layers of the same network via functional map instead of concatenationDINOv2-ViT-B/14 · TSSEPE: 30.9 -> 25.9Improved
functional map versus concatenation on sparse keypointsDINOv2-ViT-B/14 + SD · SPair-71kPCK@0.1: 57.2 -> 55.3Mixed

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