Quality Evaluation Results for Simplified Real-Scene 3D Models

A dataset containing quantitative evaluation results for real-scene 3D models simplified at rates of 0%, 20%, 40%, 60%, 80%, and 90%. The data includes image-based similarity metrics, Hausdorff distance results, positional accuracy measurements, and HCI-based evaluation scores from six evaluators. It was authored by HONG WEI and last updated on June 1, 2026.

Use Cases

• Benchmarking model simplification algorithms based on quantitative metrics like SSIM, PSNR, and Hausdorff distance.
• Analyzing the relationship between simplification rate and human-computer interaction (HCI) evaluation scores.
• Assessing the reliability of HCI-based evaluation methods using the included Fleiss’ Kappa coefficients.
• Studying positional accuracy degradation in 3D models derived from UAV photogrammetry under simplification.

Strengths

• Includes results for six distinct simplification rates (0%, 20%, 40%, 60%, 80%, 90%), enabling trend analysis.
• Incorporates multiple evaluation perspectives: image-based metrics, geometric distances, positional accuracy, and scores from six experienced human evaluators.
• Provides a measure of inter-evaluator agreement via Fleiss’ Kappa coefficients.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment for large-scale training tasks.
• The dataset size is 2.0 MB, indicating a relatively small-scale study.

Provenance

Source

figshare

Collection Method

Generated from a UAV photogrammetry reconstruction used as a reference model, with simplified versions created for evaluation.

Freshness

Last updated 2026-06-01 07:34:05; freshness should be verified.