AI researcher at Joanneum Research spanning hyperspectral image classification and multilingual speech recognition.
2
papers here
2
reviewed by Evolve
1
first-author
0
corresponding
—
works · OpenAlex
4
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-16) and cover the author's entire public record, not just the papers listed here. Reproducibility signal comes from Evolve's own reviews.
Research profile
Output—
Output metrics are not yet available: total indexed works, first-author share, and yearly production could not be aggregated reliably for this OpenAlex identity. The visible footprint — seven works, h-index 4 — is too sparse to score.
Impact—
Impact data is insufficient: citation counts, citing works, and field-normalized influence are unavailable for this record. The only signals — h-index 4, i10-index 0 — are too thin; this axis, measuring accumulated citation influence, is left unscored.
Focus0/100
The focus score of 0 reflects a highly dispersed record: the OpenAlex field distribution splits evenly (0.5/0.5) between speech processing/automatic speech recognition and hyperspectral image classification, while topic clusters add advanced ceramic materials synthesis and powder metallurgy (3 mentions each) alongside neural network applications, spectroscopy/chemometrics, and remote sensing. Such breadth, on a small seven-work record, suggests both genuine cross-domain applied work and possible identity fragmentation rather than a single sustained line.
Collaboration50/100
With a collaboration score of 50 and 11 distinct coauthors on record, Vladimir works in small, field-specific teams. The frequent-coauthor list splits along the same two research lines as the field distribution: an ASR cluster including Nathan Roll, Irene Yi, Büşra Marşan, Gabriel Stein, and Calbert Graham, and an imaging/Austrian cluster around Harald Ganster at Joanneum Research, plus collaborators such as Vianney Grenez, Pavle Padjin, and Aleksandar Nešković. Neither 2026 paper lists him as corresponding author.
Activity32/100
The activity score of 32 mirrors a sparse, episodic publishing pattern: OpenAlex counts one work in each of 2007, 2009, 2011, 2018, and 2023, with 2026 the only multi-work year at two outputs. Both recent items are arXiv preprints, giving recent3yWorks = 2. Long multi-year gaps between indexed works keep the cadence low even though the 2026 cluster shows a current, active phase concentrated in applied deep learning.
Reproducibility60/100
Reproducibility averages 60 across the two Evolve-reviewed papers but splits sharply. The Orukeet ASR preprint is rated high readiness — artefacts support re-running the FLEURS evaluation — although Evolve flags test-set tuning and train–eval overlap in the 11.01%→9.85% WER claim. The first-author hyperspectral paper is rated low: the Grad-CAM wavelength-selection study lacks PCA and random-selection baselines and reports no variance, so its conclusions rest on a thin experimental protocol.
Research summary
Vladimir Zeljković is a researcher associated with Joanneum Research in Austria; this profile is name-resolved from OpenAlex public bibliographic data. The indexed record is small but current: seven works spread across 2007–2026, with two in the last three years, both 2026 arXiv preprints. The OpenAlex field distribution splits evenly (0.5/0.5) between speech processing/automatic speech recognition and hyperspectral image classification (computer vision/XAI). The two Evolve-reviewed 2026 papers map onto that split: a first-author study applying Grad-CAM to a small Conv1D network to select key SWIR wavelengths for wood-recycling hyperspectral imaging, and a middle-author contribution to Orukeet, a multilingual ASR system using frozen fitted Gabor kernels that reportedly cuts Parakeet's FLEURS WER from 11.01% to 9.85%. Earlier topic clusters — advanced ceramic materials synthesis and powder metallurgy (three mentions each), plus spectroscopy/chemometrics and neural network applications — hint at a trajectory from materials science toward applied machine learning. Citation impact is modest (h-index 4, i10-index 0); reproducibility readiness averages 60 across the two reviewed papers, high for the ASR work and low for the hyperspectral study. Radar scores show mid-range collaboration (50), low activity (32), and a focus score of 0, consistent with a dispersed cross-domain record.
Collaboration pattern
Collaboration score 50 with 11 distinct coauthors points to small, field-specific teams rather than one large standing group. The network mirrors the dual field split: an ASR cluster (Nathan Roll, Irene Yi, Büşra Marşan, Gabriel Stein, Calbert Graham) behind the Orukeet preprint, and an Austrian imaging cluster around Joanneum Research's Harald Ganster on the hyperspectral side, with further ties to Vianney Grenez, Pavle Padjin, Branka Stojanović, Momcilo Mrkaic, and Aleksandar Nešković. He is first author on the imaging paper but not corresponding on either.
Research keywords
Highlights
Grad-CAM-guided wavelength selection for wood-recycling hyperspectral imaging
Orukeet: frozen Gabor kernels for multilingual ASR
Timeline
- 2007Earliest work in the OpenAlex record: one indexed publication, starting a pattern of single-work years.
- 2011Another single-work year; after 2011 the record goes quiet until 2018.
- 2018One indexed work appears after a seven-year gap, continuing the sparse cadence.
- 2023A single indexed work, the most recent output before the 2026 cluster.
- 2026Most productive year on record: two arXiv preprints — a first-author XAI hyperspectral study and the Orukeet multilingual ASR paper — both reviewed by Evolve.
Papers in the Evolve catalog
Frequent co-authors
About this profile
This profile is AI-structured by Evolve.Pub from public bibliographic data (OpenAlex) and Evolve's own paper reviews. It is not written or endorsed by the researcher unless the profile is claimed. Radar dimensions are computed from real bibliometrics with fixed formulas — raw figures are shown alongside; missing dimensions mean insufficient data, never zero. Errors? Claim the profile to correct or hide it.
- Bibliometrics come from OpenAlex only; no ORCID is linked, so the identity is name-resolved from public bibliographic records and may merge or split people sharing this name.
- Output and impact radar scores are not computed: only h-index 4 and i10-index 0 are available, with no aggregated publication or citation counts.
- Qualitative assessments rest on two Evolve-reviewed 2026 arXiv preprints; no earlier works were individually reviewed.
- The mix of ceramics/powder-metallurgy topics with deep-learning applications suggests possible record fragmentation, so field-level conclusions carry that uncertainty.