Nested Invariance Pooling and RBM Hashing for Image Instance Retrieval

TL;DR

This paper introduces Nested Invariance Pooling (NIP) for extracting compact, invariant image descriptors and uses RBM hashing with a novel regularization to produce highly effective binary hashes for image retrieval.

Contribution

It proposes NIP, a new method for invariant feature extraction inspired by i-theory, and a novel RBM-based hashing scheme for image retrieval.

Findings

NIP descriptors outperform existing features in retrieval tasks.

RBM hashing with regularization achieves high accuracy with very compact codes.

The combined approach is effective across diverse datasets.

Abstract

The goal of this work is the computation of very compact binary hashes for image instance retrieval. Our approach has two novel contributions. The first one is Nested Invariance Pooling (NIP), a method inspired from i-theory, a mathematical theory for computing group invariant transformations with feed-forward neural networks. NIP is able to produce compact and well-performing descriptors with visual representations extracted from convolutional neural networks. We specifically incorporate scale, translation and rotation invariances but the scheme can be extended to any arbitrary sets of transformations. We also show that using moments of increasing order throughout nesting is important. The NIP descriptors are then hashed to the target code size (32-256 bits) with a Restricted Boltzmann Machine with a novel batch-level regularization scheme specifically designed for the purpose of hashing (RBMH). A thorough empirical evaluation with state-of-the-art shows that the results obtained both with the NIP descriptors and the NIP+RBMH hashes are consistently outstanding across a wide range of datasets.