Abstract:Most image instance retrieval pipelines are based on comparison of vectors known as global image descriptors between a query image and the database images. Due to their success in large scale image classification, representations extracted from Convolutional Neural Networks (CNN) are quickly gaining ground on Fisher Vectors (FVs) as state-of-the-art global descriptors for image instance retrieval. While CNN-based descriptors are generally remarked for good retrieval performance at lower bitrates, they nevertheless present a number of drawbacks including the lack of robustness to common object transformations such as rotations compared with their interest point based FV counterparts. In this paper, we propose a method for computing invariant global descriptors from CNNs. Our method implements a recently proposed mathematical theory for invariance in a sensory cortex modeled as a feedforward neural network. The resulting global descriptors can be made invariant to multiple arbitrary transformation groups while retaining good discriminativeness. Based on a thorough empirical evaluation using several publicly available datasets, we show that our method is able to significantly and consistently improve retrieval results every time a new type of invariance is incorporated. We also show that our method which has few parameters is not prone to overfitting: improvements generalize well across datasets with different properties with regard to invariances. Finally, we show that our descriptors are able to compare favourably to other state-of-the-art compact descriptors in similar bitranges, exceeding the highest retrieval results reported in the literature on some datasets. A dedicated dimensionality reduction step --quantization or hashing-- may be able to further improve the competitiveness of the descriptors.

Using Invariant Feature in Content-Based Image Retrieval

Image Retrieval Based on RST Invariant Features Extracted from Scale Invariant Keypoints

Image Retrieval Based On Rst-Invariant Features

Texture Image Retrieval Based on Rotated Complex Wavelet Signatures.

Content-Based Image Retrieval By Dominant Color Feature

Image Retrieval Method Based Texture Watermarks

Robust Image Hashing Based on Statistical Invariance of DCT Coefficients.

Image Retrieval Based on Affine Invariant Feature in Object Regions

A Robust Affine Invariant Feature Matching Approach.

A Novel Image Signature Method for Content Authentication

A New Robust Content-Based Image Authentication Scheme

A Projective Invariant for Textures

Robust Affine Invariant Feature Extraction for Image Matching

Group Invariant Deep Representations for Image Instance Retrieval

A Textural Feature-Based Image Retrieval Algorithm

Affine invariant descriptors for color images based on independent component analysis

Affine-invariant image retrieval based on wavelet interest points

Scale and Color Invariant Image Feature Description

Invariant Description and Retrieval of Planar Shapes Using Radon Composite Features

Large Scale Partial-Duplicate Image Retrieval Using Invariance Weight of SIFT and SROA Geometric Consistency

Three-dimensional integral invariant signatures and their application on face recognition