Work

Hashing-Baseline: Rethinking Hashing in the Age of Pretrained Models

Year: 2026

Type: conference-paper

Abstract: Information retrieval with compact binary codes, 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 pre-trained encoders that produce rich 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.1

Source: IEEE International Conference on Acoustics Speech and Signal Processing

DOI: 10.1109/icassp55912.2026.11464631

Authors: Ilyass Moummad, Kawtar Zaher, Lukas Rauch, Alexis Joly

Institutions: Institut national de recherche en sciences et technologies du numérique, Laboratoire d'Informatique, de Robotique et de Microélectronique de Montpellier, University of Kassel

Language: English

Topic: Advanced Image and Video Retrieval Techniques

Subfield: Computer Vision and Pattern Recognition

Field: Computer Science

Domain: Physical Sciences

Open Access status: green

Funder: Agence Nationale de la Recherche

Awards: G1190173965, New perspectives on plant disease characterization and species associations based on deep learning and citizen science