Our Work

Segmentation masks of dental biofilm on removable dental prostheses, for research on automatic biofilm detection without disclosing agents. This record contains 33 binary annotation masks (PNG) from 1…

Segmentation masks of dental biofilm on removable dental prostheses, for research on automatic biofilm detection without disclosing agents. This record contains 33 binary annotation masks (PNG) from 11 patients. Each mask was annotated manually by a dental professional directly on the undyed photograph of the prosthesis. White (255) indicates biofilm; black (0) indicates background and clean prosthesis surface. Resolution is 6000 × 4000 pixels, matching the original photographs. Filenames follow the pattern pac_NN__view.png. Ethics: study approved by the Research Ethics Committee (CAAE No. 96996226.8.0000.5142). By determination of the Committee, the patient photographs are not publicly released. Only these annotation masks — which contain no identifiable information — and the accompanying documentation are made public. Context: the masks derive from an original, purpose-built image bank of removable dental prostheses photographed under standardized studio conditions in multiple views. For each prosthesis, several states were recorded: undyed, dyed with disclosing agent, cleaned, and manual annotation. The full bank comprises 22 patients; photographs remain restricted to the research team. Masks were annotated on the undyed image because the prosthesis is handled between acquisitions and returns to a slightly different position, preventing precise spatial alignment between the dyed and undyed photographs. Annotation is ongoing; 18 additional masks are pending and may be included in future versions.

This repository contains the code, input data, and results for the real-dataexperiments (Section 9.6, "Real-data validation on public benchmark data") ofthe paper above, submitted to Computers & S…

This repository contains the code, input data, and results for the real-dataexperiments (Section 9.6, "Real-data validation on public benchmark data") ofthe paper above, submitted to Computers & Security (Elsevier). Itaccompanies, but is independent of, the paper's constructed-scenarioexperiments (Section 9, Experiments 1-5), which require no external data andare given directly in the paper text.

Reproduction artifact for the paper "Label-Free Detection of Poisoned Dependencies: The Reachable Frontier and Its Seam"  The paper is a measurement study of how far a purely label-free, benign-o…

Reproduction artifact for the paper "Label-Free Detection of Poisoned Dependencies: The Reachable Frontier and Its Seam"  The paper is a measurement study of how far a purely label-free, benign-only one-class anomaly detector reaches against poisoned open-source packages across PyPI, npm, RubyGems, and C#/.NET, and where it fails: the detectability "seam" (the recall floor as a payload blends into ordinary library code) and the adaptive-evasion boundary. The detector fits a frozen base code encoder (ModernBERT-base) on benign packages only and operates as a review-budget router, never a gate. The artifact contains: (i) the complete one-class harness — featurization, kNN / Isolation-Forest anomaly engines, the hand-weighted capability-rule baseline, and the ablation, detectability, evasion, deployment, size-confound, and fair-capability scripts; (ii) the frozen semi-synthetic C# corpora underlying RQ4/RQ5; (iii) the per-run result JSONs backing every table and figure (RQ1–RQ6); and (iv) per-ecosystem package manifests for re-fetching the benign and malicious corpora. Everything is static — no package is executed — and the fit uses no malware labels. Malicious samples are from the Backstabber's Knife Collection (Ohm et al., 2020) and are referenced by manifest rather than redistributed. Code is MIT-licensed.

FIGURE 2. Hapalogenius beaveri sp. nov., female: dorsal habitus (Bar 0.5 mm) (A), denticulate pronotal scale-like seta (Bar 5 μm) (B), elongate denticulate and pyriform echinate pronotal scale-like se…

FIGURE 2. Hapalogenius beaveri sp. nov., female: dorsal habitus (Bar 0.5 mm) (A), denticulate pronotal scale-like seta (Bar 5 μm) (B), elongate denticulate and pyriform echinate pronotal scale-like setae (Bar 50 μm) (C), enlarged section of the elytral disc (Bar 40 μm) (D), oval echinate elytral scale-like seta (Bar 5 μm) (E), elongate denticulate elytral scale-like setae with one oval echinate scale-like seta (Bar 20 μm) (F), enlarged section of the elytral declivity with scabrous scale-like setae (Bar 40 μm) (G). Photos: Ralf Thomann.

FIGURE 1. Haplogenius beaveri sp. nov.: male, dorsal habitus (Bar 0.2 mm) (A); male, frons anterior (Bar 0.1 mm) (B); female, lateral view (Bar 0.1 mm) (C); male, elytron (Bar 0.1 mm) (D); male, elytr…

FIGURE 1. Haplogenius beaveri sp. nov.: male, dorsal habitus (Bar 0.2 mm) (A); male, frons anterior (Bar 0.1 mm) (B); female, lateral view (Bar 0.1 mm) (C); male, elytron (Bar 0.1 mm) (D); male, elytral scale-like setae (Bar 0.05 mm) (E); male, declivity (Bar 0.1 mm) (F). Photos: Christoph Allgaier.

Gebhardt, Heiko (2026): A new species of the genus Hapalogenius Hagedorn with a detailed description of the microsculpture of its squamose vestiture (Coleoptera: Curculionidae). Zootaxa 5853 (4): 593-…

Gebhardt, Heiko (2026): A new species of the genus Hapalogenius Hagedorn with a detailed description of the microsculpture of its squamose vestiture (Coleoptera: Curculionidae). Zootaxa 5853 (4): 593-600, DOI: 10.11646/zootaxa.5853.4.9, URL: https://doi.org/10.11646/zootaxa.5853.4.9

FIGURES 21–22. Glans penis of Dinaric Nemaspela species: 21. Nemaspela ladae Karaman, 2013, dorsal view, redrawn from SEM image in Karaman (2013); 22. Nemaspela borkoae Kozel, Delić & Novak, 2020,…

FIGURES 21–22. Glans penis of Dinaric Nemaspela species: 21. Nemaspela ladae Karaman, 2013, dorsal view, redrawn from SEM image in Karaman (2013); 22. Nemaspela borkoae Kozel, Delić & Novak, 2020, ventral view, redrawn from image in Kozel et al. 2020.