Collision-Resistant Single-Pass Method for Unsupervised Fine-Grained Image Hashing

TL;DR

This paper introduces CS3H, a novel unsupervised image hashing method that enhances collision resistance and fine-grained discrimination using a single-pass loss and attention module, outperforming existing methods.

Contribution

The paper proposes a collision-resistant framework with a new loss function and attention module for unsupervised fine-grained image hashing, improving retrieval accuracy and collision resistance.

Findings

CS3H outperforms state-of-the-art methods in retrieval accuracy.

CS3H achieves superior collision resistance with minimal computational overhead.

The collision-sensitive attention module effectively emphasizes rare local patterns.

Abstract

Unsupervised fine-grained image hashing aims to learn compact binary codes that preserve subtle visual differences among highly similar instances without manual annotations. However, most existing methods neglect collision resistance, leading to identical hash codes for slightly semantically different samples. In this paper, we propose Collision-Resistant Single-Pass Self-Supervised Semantic Hashing (CS3H), a collision-resistant framework that directly optimizes Hamming-space similarity via a single-pass normalized Hamming distance loss to produce well-separated binary representations. We further introduce a collision-sensitive attention module to emphasize rare and discriminative local patterns, reducing hash collisions and improving fine-grained discrimination. Experiments on multiple benchmarks show that CS3H consistently outperforms state-of-the-art methods in retrieval accuracy while achieving superior collision resistance with minimal computational overhead.