Deep Hash Distillation For Image Retrieval

Abstract

In hash-based image retrieval systems, degraded or transformed inputs usually generate different codes from the original, deteriorating the retrieval accuracy. To mitigate this issue, data augmentation can be applied during training. However, even if augmented samples of an image are similar in real feature space, the quantization can scatter them far away in Hamming space. This results in representation discrepancies that can impede training and degrade performance. In this work, we propose a novel self-distilled hashing scheme to minimize the discrepancy while exploiting the potential of augmented data. By transferring the hash knowledge of the weakly-transformed samples to the strong ones, we make the hash code insensitive to various transformations. We also introduce hash proxy-based similarity learning and binary cross entropy-based quantization loss to provide fine quality hash codes. Ultimately, we construct a deep hashing framework that not only improves the existing deep hashi

Stats

• citations31
• S2 citations—
• github stars0
• HF likes0
• heat score11.29
• arxiv keyjang2021deep

Related papers

• Deep Self-adaptive Hashing For Image Retrieval (2021)8.35
• Transhash: Transformer-based Hamming Hashing For Efficient Image Retrieval (2021)13.44
• Deep Hashing: A Joint Approach For Image Signature Learning (2016)5.24
• Query-adaptive Image Retrieval By Deep Weighted Hashing (2016)12.68
• SSDH: Semi-supervised Deep Hashing For Large Scale Image Retrieval (2016)15.40
• Transfer Adversarial Hashing For Hamming Space Retrieval (2017)8.60
• Deep Supervised Discrete Hashing (2017)0.00
• One Loss For Quantization: Deep Hashing With Discrete Wasserstein Distributional Matching (2022)12.40