N
Noe Pion
Publications - 8
Citations - 756
Noe Pion is an academic researcher. The author has contributed to research in topics: Image retrieval & Pose. The author has an hindex of 5, co-authored 7 publications receiving 292 citations.
Papers
More filters
Proceedings Article
Hard Negative Mixing for Contrastive Learning
TL;DR: It is argued that an important aspect of contrastive learning, i.e., the effect of hard negatives, has so far been neglected and proposed hard negative mixing strategies at the feature level, that can be computed on-the-fly with a minimal computational overhead.
Posted Content
R2D2: Repeatable and Reliable Detector and Descriptor.
Jerome Revaud,Philippe Weinzaepfel,César Roberto de Souza,Noe Pion,Gabriela Csurka,Yohann Cabon,Martin Humenberger +6 more
TL;DR: This work argues that salient regions are not necessarily discriminative, and therefore can harm the performance of the description, and proposes to jointly learn keypoint detection and description together with a predictor of the local descriptor discriminativeness.
Posted Content
Robust Image Retrieval-based Visual Localization using Kapture
Martin Humenberger,Yohann Cabon,Nicolas Guérin,Julien Morat,Jerome Revaud,Philippe Rerole,Noe Pion,César Roberto de Souza,Vincent Leroy,Gabriela Csurka +9 more
TL;DR: This paper presents kapture, a flexible data format and processing pipeline for structure from motion and visual localization that is released open source that is based on robust image retrieval for coarse camera pose estimation and robust local features for accurate pose refinement.
Posted Content
Hard Negative Mixing for Contrastive Learning
TL;DR: Zhang et al. as mentioned in this paper proposed hard negative mixing strategies at the feature level, which can be computed on-the-fly with a minimal computational overhead. And they exhaustively ablate their approach on linear classification, object detection and instance segmentation.
Posted Content
Benchmarking Image Retrieval for Visual Localization
TL;DR: It is shown that retrieval performance on classical landmark retrieval/recognition tasks correlates only for some but not all tasks to localization performance, indicating a need for retrieval approaches specifically designed for localization tasks.