Szczegóły publikacji
Opis bibliograficzny
Hyper-DINO: efficient hyperbolic embeddings for histopathological content-based image retrieval / Stanisław ŁAŻEWSKI, Bogusław CYGANEK // W: Computational Science – ICCS 2026 workshops : 26th International Conference : Hamburg, Germany, June 29–July 1, 2026 : proceedings , Pt. 2 / eds. Maciej Paszynski, Amanda S. Barnard, Yongjie Jessica Zhang. — Cham : Springer, cop. 2026. — ( Lecture Notes in Computer Science ; ISSN 0302-9743 ; LNCS 16787 ). — ISBN: 978-3-032-29908-6; e-ISBN: 978-3-032-29909-3. — S. 473–487. — Bibliogr., Abstr. — Publikacja dostępna online od: 2026-06-28
Autorzy (2)
Słowa kluczowe
Dane bibliometryczne
| ID BaDAP | 168927 |
|---|---|
| Data dodania do BaDAP | 2026-08-31 |
| DOI | 10.1007/978-3-032-29909-3_34 |
| Rok publikacji | 2026 |
| Typ publikacji | materiały konferencyjne (aut.) |
| Otwarty dostęp |  |
| Wydawca | Springer |
| Konferencja | International Conference on Computational Science 2026 |
| Czasopismo/seria | Lecture Notes in Computer Science |
Abstract
In this paper, we propose new hyperbolic embeddings, Hyper-DINO, for use in histopathological content-based image retrieval. First, we used a DINO pretrained ViT Small vision transformer to extract the base features. The resulting [CLS] token is dimensionally reduced using PCA or NCA techniques. These features are then mapped onto a hyperbolic space, so that features from positive pairs are closer together, while features from negative pairs are further apart. For this mapping, we use a specially designed Hyperbolic Mapper model with Hyperbolic Contrastive Loss function. This approach improved MAP@20 results on the Kather and BreakHis histopathology datasets at separate magnifications of 40X, 100X, 200X, and 400X, respectively, to 94.61%, 86.64%, 79.15%, 78.91%, and 75.17%. An additional advantage of the Hyper-DINO features is their compact size. The representation of a single image is a vector of 32 float16 numbers, so it takes up only 64 bytes.
