用预训练模型直接生成哈希码，无需训练即可实现高效检索。
Hashing-Baseline: Rethinking Hashing in the Age of Pretrained Models
- 结合主成分分析、随机正交投影和阈值二值化，直接生成二进制哈希码。
- 在图像与音频检索任务上达到与训练型方法相当的性能。
- 适用于需要快速部署的场景，尤其适合资源受限环境。
基于紧凑二进制嵌入的信息检索（即哈希）对可扩展的快速搜索应用至关重要，但现有先进哈希方法需昂贵且场景特定的训练。本文提出 Hashing-Baseline，一种无需训练的强基线哈希方法，利用强大的预训练编码器生成丰富的预训练嵌入。我们重新审视经典无训练哈希技术：主成分分析、随机正交投影和阈值二值化，构建哈希强基线。该方法将这些技术与最先进的视觉和音频编码器的冻结嵌入结合，在无需额外学习或微调的情况下，实现具有竞争力的检索性能。为验证其通用性与有效性，我们在标准图像检索基准以及新提出的音频哈希基准上进行了评估。
原文摘要 · Abstract (English)
Information retrieval with compact binary embeddings, also referred to as hashing, is crucial for scalable fast search applications, yet state-of-the-art hashing methods require expensive, scenario-specific training. In this work, we introduce Hashing-Baseline, a strong training-free hashing method leveraging powerful pretrained encoders that produce rich pretrained embeddings. We revisit classical, training-free hashing techniques: principal component analysis, random orthogonal projection, and threshold binarization, to produce a strong baseline for hashing. Our approach combines these techniques with frozen embeddings from state-of-the-art vision and audio encoders to yield competitive retrieval performance without any additional learning or fine-tuning. To demonstrate the generality and effectiveness of this approach, we evaluate it on standard image retrieval benchmarks as well as a newly introduced benchmark for audio hashing.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。