Rescuing Deep Hashing from Dead Bits Problem

02/01/2021
by   Shu Zhao, et al.
0

Deep hashing methods have shown great retrieval accuracy and efficiency in large-scale image retrieval. How to optimize discrete hash bits is always the focus in deep hashing methods. A common strategy in these methods is to adopt an activation function, e.g. sigmoid(·) or tanh(·), and minimize a quantization loss to approximate discrete values. However, this paradigm may make more and more hash bits stuck into the wrong saturated area of the activation functions and never escaped. We call this problem "Dead Bits Problem (DBP)". Besides, the existing quantization loss will aggravate DBP as well. In this paper, we propose a simple but effective gradient amplifier which acts before activation functions to alleviate DBP. Moreover, we devise an error-aware quantization loss to further alleviate DBP. It avoids the negative effect of quantization loss based on the similarity between two images. The proposed gradient amplifier and error-aware quantization loss are compatible with a variety of deep hashing methods. Experimental results on three datasets demonstrate the efficiency of the proposed gradient amplifier and the error-aware quantization loss.

READ FULL TEXT
research
10/24/2021

Deep Asymmetric Hashing with Dual Semantic Regression and Class Structure Quantization

Recently, deep hashing methods have been widely used in image retrieval ...
research
10/31/2017

Deep Hashing with Triplet Quantization Loss

With the explosive growth of image databases, deep hashing, which learns...
research
10/30/2016

Accurate Deep Representation Quantization with Gradient Snapping Layer for Similarity Search

Recent advance of large scale similarity search involves using deeply le...
research
11/02/2017

Set-to-Set Hashing with Applications in Visual Recognition

Visual data, such as an image or a sequence of video frames, is often na...
research
10/05/2022

Active Image Indexing

Image copy detection and retrieval from large databases leverage two com...
research
09/17/2020

Deep Momentum Uncertainty Hashing

Discrete optimization is one of the most intractable problems in deep ha...
research
10/01/2021

Beyond Neighbourhood-Preserving Transformations for Quantization-Based Unsupervised Hashing

An effective unsupervised hashing algorithm leads to compact binary code...

Please sign up or login with your details

Forgot password? Click here to reset