Academic Journal
From pixels to perception: A benchmark for human-like symmetry detection.
| Title: | From pixels to perception: A benchmark for human-like symmetry detection. |
|---|---|
| Authors: | Muradás Odriozola G; Department of Brain and Cognition, University of Leuven (KU Leuven), Tiensestraat 102 - Box 3711, Leuven, 3000, Belgium; Image and Speech Processing (PSI), Department of Electrical Engineering (ESAT), Castle Park Arenberg 10 - bus 2440, Leuven, 3001, Belgium; Leuven.AI, KU Leuven Institute for AI, Leuven, 3000, Belgium. Electronic address: gonzalo.muradasodriozola@kuleuven.be., Koßmann L; Department of Brain and Cognition, University of Leuven (KU Leuven), Tiensestraat 102 - Box 3711, Leuven, 3000, Belgium; Leuven.AI, KU Leuven Institute for AI, Leuven, 3000, Belgium., Tuytelaars T; Image and Speech Processing (PSI), Department of Electrical Engineering (ESAT), Castle Park Arenberg 10 - bus 2440, Leuven, 3001, Belgium; Leuven.AI, KU Leuven Institute for AI, Leuven, 3000, Belgium., Wagemans J; Department of Brain and Cognition, University of Leuven (KU Leuven), Tiensestraat 102 - Box 3711, Leuven, 3000, Belgium; Leuven.AI, KU Leuven Institute for AI, Leuven, 3000, Belgium. |
| Source: | Vision research [Vision Res] 2026 Aug; Vol. 245, pp. 108825. Date of Electronic Publication: 2026 May 05. |
| Publication Type: | Journal Article |
| Language: | English |
| Journal Info: | Publisher: Elsevier Science Ltd Country of Publication: England NLM ID: 0417402 Publication Model: Print-Electronic Cited Medium: Internet ISSN: 1878-5646 (Electronic) Linking ISSN: 00426989 NLM ISO Abbreviation: Vision Res Subsets: MEDLINE |
| Imprint Name(s): | Publication: Kidlington, Oxford : Elsevier Science Ltd. Original Publication: Oxford [etc.] |
| MeSH Terms: | Form Perception*/physiology , Pattern Recognition, Visual*/physiology , Visual Perception*/physiology , Detection Algorithms*, Humans ; Benchmarking |
| Abstract: | Symmetry, a fundamental concept in nature, science and art, has challenged computer vision researchers because it occurs in various forms and human symmetry perception can deviate from the mathematical definition. Previous symmetry detection datasets are limited by the number of annotators and by missing the nuances of human perception. We introduce PIX2PER, a novel dataset for reflection symmetry in natural scenes and artworks. We also introduce WF1, a modified version of the widely-used F1 detection performance score, by adding weights to precision and recall to accommodate for the perceived symmetry strength. Created by adding weights to precision and recall to accommodate for the perceived symmetry strength. We perform a comparative analysis of existing models for symmetry detection on this human-centric dataset. Additionally, we present a fully synthetic dataset for pretraining symmetry detection models. When finetuning this pretrained model with human data, performance increases significantly. This research introduces and evaluates ways of improving symmetry detection and contributes to the development of computer vision models that more effectively represent human perception. (Copyright © 2026 The Authors. Published by Elsevier Ltd.. All rights reserved.) |
| References: | Spat Vis. 1995;9(1):9-32. (PMID: 7626549) J Exp Psychol Hum Percept Perform. 1993 Apr;19(2):364-80. (PMID: 8473845) Vis Neurosci. 1999 Sep-Oct;16(5):919-22. (PMID: 10580727) Perception. 1993;22(5):565-87. (PMID: 8414882) Percept Psychophys. 1991 Nov;50(5):413-27. (PMID: 1788030) J Vis. 2001;1(2):137-44. (PMID: 12678608) IEEE Trans Image Process. 2021;30:5708-5723. (PMID: 34138706) Atten Percept Psychophys. 2010 Oct;72(7):1854-64. (PMID: 20952783) J Vis. 2009 Nov 18;9(12):11.1-11. (PMID: 20053102) J Exp Psychol Hum Percept Perform. 1981 Dec;7(6):1186-210. (PMID: 6458647) Neural Netw. 2005 Jan;18(1):23-32. (PMID: 15649659) PeerJ. 2019 Jun 17;7:e7078. (PMID: 31245176) J Vis. 2008 May 27;8(5):14.1-18. (PMID: 18842085) Perception. 2007;36(9):1305-19. (PMID: 18196698) PeerJ. 2019 Dec 12;7:e8169. (PMID: 31844575) Acta Psychol (Amst). 2004 Nov;117(3):233-50. (PMID: 15500805) J Vis. 2014 Mar 11;14(3):14. (PMID: 24618107) Vision Res. 2001 Nov;41(24):3163-83. (PMID: 11711141) |
| Contributed Indexing: | Keywords: Art images; Benchmark dataset; Computer vision; Human symmetry perception; Natural images; Synthetic data |
| Entry Date(s): | Date Created: 20260506 Date Completed: 20260613 Latest Revision: 20260813 |
| Update Code: | 20260813 |
| PubMed Central ID: | PMC13282018 |
| DOI: | 10.1016/j.visres.2026.108825 |
| PMID: | 42090885 |
| Database: | MEDLINE |
| ISSN: | 1878-5646 |
|---|---|
| DOI: | 10.1016/j.visres.2026.108825 |