Learning Super-features For Image Retrieval

Abstract

Methods that combine local and global features have recently shown excellent performance on multiple challenging deep image retrieval benchmarks, but their use of local features raises at least two issues. First, these local features simply boil down to the localized map activations of a neural network, and hence can be extremely redundant. Second, they are typically trained with a global loss that only acts on top of an aggregation of local features; by contrast, testing is based on local feature matching, which creates a discrepancy between training and testing. In this paper, we propose a novel architecture for deep image retrieval, based solely on mid-level features that we call Super-features. These Super-features are constructed by an iterative attention module and constitute an ordered set in which each element focuses on a localized and discriminant image pattern. For training, they require only image labels. A contrastive loss operates directly at the level of Super-features a

Stats

• citations0
• S2 citations—
• github stars142
• HF likes0
• heat score4.31
• arxiv keyweinzaepfel2022learning

Related papers

• Global Features Are All You Need For Image Retrieval And Reranking (2023)17.53
• Unifying Deep Local And Global Features For Image Search (2020)28.10
• DALG: Deep Attentive Local And Global Modeling For Image Retrieval (2022)0.00
• Deep Image Retrieval: Learning Global Representations For Image Search (2016)19.67
• Supergf: Unifying Local And Global Features For Visual Localization (2022)0.00
• Global-to-local Or Local-to-global? Enhancing Image Retrieval With Efficient Local Search And Effective Global Re-ranking (2025)0.00
• Aggregated Deep Local Features For Remote Sensing Image Retrieval (2019)14.11
• Local Features And Visual Words Emerge In Activations (2019)13.55