Research

Paper

TESTING March 24, 2026

Knot-10:A Tightness-Stratified Benchmark for Real-World Knot Classification with Topological Difficulty Analysis

Authors

Shiheng Nie, Yunguang Yue

Abstract

Physical knot classification is a fine-grained visual classification (FGVC) scenario in which appearance cues are deliberately suppressed: different classes share the same rope material, color, and background, and class identity resides primarily in crossing structure. We introduce the Knots-10 benchmark, comprising 1,440 images with a deployment-oriented split that trains on loosely tied knots and tests on tightly dressed ones. Swin-T and TransFG both average 97.2% accuracy; PMG scores 94.5%, consistent with the hypothesis that jigsaw shuffling disrupts crossing continuity. McNemar tests cannot separate four of the five general-purpose backbones, so small ranking margins should be interpreted with caution. A Mantel permutation test shows that topological distance significantly correlates with confusion patterns in three of the five models (p < 0.01). We propose TACA regularization, which improves embedding-topology alignment from rho=0.46 to rho=0.65 without improving classification accuracy; a random-distance ablation yields comparable alignment, indicating the benefit is likely driven by generic regularization. A pilot cross-domain test with 100 phone photographs reveals a 58-69 percentage-point accuracy drop, exposing rope appearance bias as the dominant failure mode.

Metadata

arXiv ID: 2603.23286
Provider: ARXIV
Primary Category: cs.CV
Published: 2026-03-24
Fetched: 2026-03-25 06:02

Related papers

Raw Data (Debug)
{
  "raw_xml": "<entry>\n    <id>http://arxiv.org/abs/2603.23286v1</id>\n    <title>Knot-10:A Tightness-Stratified Benchmark for Real-World Knot Classification with Topological Difficulty Analysis</title>\n    <updated>2026-03-24T14:50:34Z</updated>\n    <link href='https://arxiv.org/abs/2603.23286v1' rel='alternate' type='text/html'/>\n    <link href='https://arxiv.org/pdf/2603.23286v1' rel='related' title='pdf' type='application/pdf'/>\n    <summary>Physical knot classification is a fine-grained visual classification (FGVC) scenario in which appearance cues are deliberately suppressed: different classes share the same rope material, color, and background, and class identity resides primarily in crossing structure. We introduce the Knots-10 benchmark, comprising 1,440 images with a deployment-oriented split that trains on loosely tied knots and tests on tightly dressed ones. Swin-T and TransFG both average 97.2% accuracy; PMG scores 94.5%, consistent with the hypothesis that jigsaw shuffling disrupts crossing continuity. McNemar tests cannot separate four of the five general-purpose backbones, so small ranking margins should be interpreted with caution. A Mantel permutation test shows that topological distance significantly correlates with confusion patterns in three of the five models (p &lt; 0.01). We propose TACA regularization, which improves embedding-topology alignment from rho=0.46 to rho=0.65 without improving classification accuracy; a random-distance ablation yields comparable alignment, indicating the benefit is likely driven by generic regularization. A pilot cross-domain test with 100 phone photographs reveals a 58-69 percentage-point accuracy drop, exposing rope appearance bias as the dominant failure mode.</summary>\n    <category scheme='http://arxiv.org/schemas/atom' term='cs.CV'/>\n    <published>2026-03-24T14:50:34Z</published>\n    <arxiv:comment>48 pages, 12 figures, 10 supplementary sections</arxiv:comment>\n    <arxiv:primary_category term='cs.CV'/>\n    <author>\n      <name>Shiheng Nie</name>\n    </author>\n    <author>\n      <name>Yunguang Yue</name>\n    </author>\n  </entry>"
}