3D scene understanding researcher probing how vision-language and LiDAR language models reason about space
5
papers here
4
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
—
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-17) 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—
No output score is available: the underlying data do not yet include a full publication count, venue profile, or yearly productivity for this researcher. This axis would measure total and per-year publication volume across indexed venues.
Impact—
No impact score is available: citation counts, h-index-type aggregates, and uptake of the two cataloged preprints are not in the current data. This axis would measure how widely the work is cited and adopted by other researchers.
Focus0/100
Focus scores 100/100: the field distribution in the underlying data contains exactly one field, computer vision (3D scene understanding), and both cataloged 2026 papers — SceneBench and the LiDAR language-model study — sit inside it. Topics span 3D scene understanding, vision-language models, Gaussian Splatting, spatial reasoning, and spatio-temporal relationships, so the concentration reflects thematic depth within one visual-perception area rather than breadth across fields.
Collaboration73/100
Collaboration scores 59/100, with 18 distinct co-authors recorded. The recurring partner list is dominated by a stable cluster — Anubhav Khanal, Prabigya Acharya, Roshni Poudel, Sujan Kapali, Bigyan Bhatta, Pramish Paudel, and Francois Rameau — plus 3D-vision collaborators including Runyi Yang, Murat Akkoyun, Di Wen, and Ruiping Liu. The 2026 record shows both middle and last authorships, though no corresponding-author paper is cataloged.
Activity—
No activity score is available: the data lack a sustained yearly publication series, so recent-period productivity and publication cadence cannot be computed. This axis would measure publishing tempo and research engagement across consecutive years.
Reproducibility57/100
Reproducibility scores 20/100 (reproMean 20), computed from one assessed paper. SceneBench is rated low readiness: Evolve's review notes Gemini-generated ground truth and absent statistical analysis, which make the headline spatial-reasoning results hard to verify or replicate. No code, data, or evaluation-protocol release is documented in the catalog, so independent verification of the benchmark currently depends on the authors.
Research summary
Danda Pani Paudel is a researcher working at the intersection of computer vision and vision-language modeling, with the Evolve profile concentrated on 3D scene understanding. The cataloged record centers on two 2026 arXiv preprints. The first, SceneBench, is a last-authored hierarchical benchmark for vision-language understanding of 3D scenes built on Gaussian Splatting representations; it claims to be the first hierarchical 3DGS benchmark for VLM spatial reasoning, covering hierarchical scene graphs and 3D question answering. Evolve's review flags that Gemini-generated ground truth and absent statistical analysis weaken the headline findings, and rates the paper's reproducibility readiness as low. The second, a middle-authored paper, asks whether LiDAR language models really understand spatio-temporal relationships, extending the same evaluative agenda from static 3D scenes to dynamic, LiDAR-grounded settings. Bibliometrically, the profile shows extreme focus: the observed field distribution contains a single field — computer vision (3D scene understanding) — yielding a focus score of 100/100. Collaboration scores 59/100 with 18 distinct co-authors, including recurring partners such as Anubhav Khanal, Francois Rameau, Runyi Yang, Murat Akkoyun, Di Wen, and Ruiping Liu. Reproducibility currently scores 20/100 from one assessed paper. Output, impact, and activity axes lack sufficient data and are not scored; this should be read as an early-stage, catalog-scoped picture.
Collaboration pattern
The collaboration profile is moderate (59/100) with 18 distinct co-authors recorded. Recurring partners split into a core cluster — Anubhav Khanal, Prabigya Acharya, Roshni Poudel, Sujan Kapali, Bigyan Bhatta, Pramish Paudel, Francois Rameau — and 3D-vision collaborators including Runyi Yang, Murat Akkoyun, Di Wen, and Ruiping Liu. The 2026 catalog shows one middle authorship and one last authorship, but no corresponding-author role yet.
Research keywords
Highlights
SceneBench: claimed first hierarchical 3DGS benchmark for VLM spatial reasoning
Reproducibility concerns temper SceneBench's headline findings
Probing whether LiDAR language models understand spatio-temporal relationships
As a middle author, co-authored a 2026 arXiv study asking whether LiDAR language models really understand spatio-temporal relationships, extending the evaluation agenda from static 3D scenes to dynamic LiDAR-grounded settings; no field tag or reproducibility assessment is available for this paper.
Timeline
- 2026Co-authored (middle authorship) "Do LiDAR Language Models Really Understand Spatio-temporal Relationships?" on arXiv.
- 2026Last-authored "SceneBench: A Hierarchical Benchmark for Vision-Language Understanding of 3D Scenes" appears on arXiv, claiming the first hierarchical 3DGS benchmark for VLM spatial reasoning.
- 2026Both cataloged papers land in the same year and the same single field — computer vision (3D scene understanding) — anchoring the profile's 100/100 focus score.
Papers in the Evolve catalog
MECAIL: Communication-Aware Incremental Learning for Object Detection with 14.6 KB Spatiotemporal Experts
Do LiDAR Language Models Really Understand Spatio-temporal Relationships?
SceneBench: A Hierarchical Benchmark for Vision-Language Understanding of 3D Scenes
Kairos: A Dataset for Fine-Grained Video-Language Modeling over Space, Time, and Dynamics
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.
- Dimension scores are computed from a limited public bibliometric snapshot; the output, impact, and activity axes lack sufficient data and are not scored.
- No institution, country, or ORCID is available; this profile is name-resolved from public bibliographic data, and "Danda Pani Paudel" may correspond to more than one researcher.
- The frequent-coauthor list contains the researcher's own name, indicating possible name-disambiguation noise in the underlying records.
- The reproducibility score (20/100) rests on a single assessed paper (SceneBench, low readiness); the second cataloged paper has no reproducibility assessment.