UIUC researcher bridging computer vision and robot learning — multimodal models, video editing, tactile manipulation.
2
papers here
2
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
17
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—
Not yet scored: this snapshot lacks a complete venue-resolved publication count, so overall output cannot be computed. The axis would measure total production volume and its spread across venues; only partial signals such as recent3yWorks (66) are currently available.
Impact—
Not yet scored: field-normalized citation data required for impact is unavailable. The axis would measure scientific influence through citation reach and field percentiles; the snapshot lists h-index 17 and i10-index 26 but derives no score from them.
Focus0/100
This axis scores 0, reflecting maximal breadth rather than inactivity: the field distribution is an exact 50/50 split between computer vision and robot learning (dexterous manipulation). Topic data confirms the spread — 41 works in multimodal ML applications alongside 20 in topic modeling, 13 in GANs/image synthesis, 11 in NLP techniques, and 11 in robot manipulation and learning. The record shows two parallel research programs rather than one narrow specialization.
Collaboration54/100
Scores 54, indicating moderate collaborative consolidation. The window records 14 distinct coauthors, and twelve recur frequently — including Adheesh Sunil Juvekar, Onkar Kishor Susladkar, Kiet A. Nguyen, and Vedant Shah — suggesting a stable core team. Both reviewed 2026 papers are multi-author efforts in which she holds middle (DexTacWAM) or last (video editing) positions, consistent with a supervisory role on vision and robotics projects rather than large-scale cross-institution networks.
Activity100/100
Scores 100, the strongest signal in the profile. Annual works climbed from 1–4 per year across 2012–2020 to 11 in 2021, 10 in 2022, 9 each in 2023–2024, then surged to 28 in 2025 and 29 in 2026, with 66 works in the trailing three-year window. This late-stage acceleration — roughly tripling from 9 works in 2024 to 28 in 2025 — is the clearest quantitative trend in the record.
Reproducibility38/100
Scores 38, computed as the mean over the two Evolve-reviewed papers (reproMean 38, reproPapers 2). DexTacWAM rates low: its gains on six real dexterous tasks are hardware-bound and single-lab, limiting independent replication. The training-free video editing framework rates medium — training-free pipelines are easier to rerun, but Evolve notes tiny ablation subsets and a strong backbone temper the headline FiVE-Acc gain of 78.16 versus 58.95. Both are arXiv preprints.
Research summary
Ismini Lourentzou is a researcher at the University of Illinois Urbana-Champaign whose record splits evenly — 50/50 in field distribution — between computer vision and robot learning for dexterous manipulation. Her OpenAlex topic profile is led by multimodal machine learning applications (41 works), followed by topic modeling (20), GANs and image synthesis (13), natural language processing techniques (11), and robot manipulation and learning (11). The bibliometric series shows a steep productivity ramp: one to four works per year from 2012 through 2020, 11 in 2021, then 28 in 2025 and 29 in 2026, with 66 works in the trailing three years — the basis for a maximal activity score. Citation aggregates record an h-index of 17 and an i10-index of 26. Evolve has reviewed two 2026 arXiv papers: DexTacWAM, a visuo-tactile world-action model whose tactile world modeling drives gains on six real dexterous tasks (low reproducibility readiness), and a unified training-free framework for diverse video editing reporting FiVE-Acc of 78.16 versus 58.95 for the best baseline (medium readiness). She works with a compact set of 14 distinct coauthors, holding middle and last author positions on the reviewed papers.
Collaboration pattern
Collaboration is compact and recurring: 14 distinct coauthors in the recorded window, with a dozen frequent collaborators — Adheesh Sunil Juvekar, Onkar Kishor Susladkar, Kiet A. Nguyen, Muntasir Wahed, Vedant Shah, and others — indicating a stable lab-style core team. On both reviewed 2026 papers she sits in middle (DexTacWAM) and last (video editing) author positions without corresponding-author flags, consistent with supervising two parallel project lines in vision and dexterous manipulation.
Research keywords
Highlights
Visuo-tactile world-action model for dexterous manipulation
Unified training-free framework for diverse video editing
Sharp publication acceleration in 2025–2026
Recorded annual works jump from 9 in 2024 to 28 in 2025 and 29 in 2026, with 66 works over the trailing three years — spanning both computer vision and dexterous manipulation output.
Timeline
- 2012Earliest recorded publication year in the OpenAlex counts-by-year series (1 work), with single-digit annual output continuing through 2017.
- 2018Annual output steps up to 4 works, beginning a gradual ramp through 2020 (3 works in 2019 and 2020).
- 2021A clear breakout year: 11 recorded works, up from 3 in 2020, followed by 10 in 2022 and 9 in each of 2023–2024.
- 2025Publication surge: 28 recorded works, roughly tripling the 2024 count of 9.
- 2026Two Evolve-reviewed arXiv papers appear — DexTacWAM in dexterous manipulation and a unified training-free video editing framework — within a record 29-work year.
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 are derived from OpenAlex only (identity: openalex); output and impact axes lack the normalized data needed for scoring.
- Reproducibility score (mean 38) is based on only two Evolve-reviewed papers, so it is a small-sample signal.
- h-index (17) and i10-index (26) are snapshot aggregates and are not field-normalized.
- 2025–2026 work counts include very recent (likely preprint-heavy) indexing and may shift as records consolidate.