Signals 4 · free daily AI digest

RePair: Turning Retrieval Failures into Counterfactual Hard Pairs

Paper recorded by Signals 4 on 2026-08-30 in cs.CV. Abstract reproduced from arXiv; link to the original below.

Published 2026-08-30 on arXiv · recorded by Signals 4 on 2026-09-01

Category: cs.CV · 计算机视觉 · first seen 2026-09-01

Abstract

Vision-language retrieval with CLIP-style dual encoders achieves strong cross-modal performance, yet practical accuracy often hinges on localized semantic distinctions where top-ranked near misses differ from the true match by a single critical detail. Hard-sample mining can select confusable candidates but cannot construct corrected counterparts; synthetic augmentation can generate novel samples

Read on arXiv →

#211 most recent of 237 cs.CV papers we have recorded · ↑ newer: SnapBench: Benchmarking Snap-and-Ask Multimodal Retrieval for Mobile I · ↓ older: Guardrail-Agnostic Societal Bias Evaluation in Large Vision-Language M
Cite this page: RePair: Turning Retrieval Failures into Counterfactual Hard Pairs: the #211 most recent of 237 cs.CV papers we have recorded (as of 2026-08-30). Source: Signals 4 (Signals API) — https://data.jiangzhang.ca/signals4/t/papers/repair-turning-retrieval-failures-into-counterfactual-hard-pairs.html
Free to quote with attribution to “Signals 4 (Signals API)”. Machine-readable: papers.json
Related: More cs.CV papers · arXiv signals · All papers · Today in AI
Get 4 AI signals a day by email — free.
Subscribe free → See all plans →
Get 4 AI signals a day by email — free
All models · All repos · By company · Daily editions