Local Features and Visual Words Emerge in Activations

We propose a novel method of deep spatial matching (DSM) for image retrieval. Initial ranking is based on image descriptors extracted from convolutional neural network activations by global pooling, as in recent state-of-the-art work. However, the same sparse 3D activation tensor is also approximate...

Full description

Saved in:
Bibliographic Details
Published in:arXiv.org 2019-05
Main Authors: Siméoni, Oriane, Avrithis, Yannis, Chum, Ondrej
Format: Article
Language:English
Subjects:
Artificial neural networks
Feature extraction
Image detection
Image management
Image retrieval
Mathematical analysis
Tensors
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:We propose a novel method of deep spatial matching (DSM) for image retrieval. Initial ranking is based on image descriptors extracted from convolutional neural network activations by global pooling, as in recent state-of-the-art work. However, the same sparse 3D activation tensor is also approximated by a collection of local features. These local features are then robustly matched to approximate the optimal alignment of the tensors. This happens without any network modification, additional layers or training. No local feature detection happens on the original image. No local feature descriptors and no visual vocabulary are needed throughout the whole process. We experimentally show that the proposed method achieves the state-of-the-art performance on standard benchmarks across different network architectures and different global pooling methods. The highest gain in performance is achieved when diffusion on the nearest-neighbor graph of global descriptors is initiated from spatially verified images.
ISSN:2331-8422