001048850 001__ 1048850
001048850 005__ 20251217202227.0
001048850 0247_ $$2datacite_doi$$a10.34734/FZJ-2025-04955
001048850 037__ $$aFZJ-2025-04955
001048850 1001_ $$0P:(DE-Juel1)190396$$aWang, Qin$$b0$$eCorresponding author$$ufzj
001048850 245__ $$aSelf-Supervised Learning based on Transformed Image Reconstruction for Equivariance-Coherent Feature Representation
001048850 260__ $$c2025
001048850 3367_ $$0PUB:(DE-HGF)25$$2PUB:(DE-HGF)$$aPreprint$$bpreprint$$mpreprint$$s1765973165_4232
001048850 3367_ $$2ORCID$$aWORKING_PAPER
001048850 3367_ $$028$$2EndNote$$aElectronic Article
001048850 3367_ $$2DRIVER$$apreprint
001048850 3367_ $$2BibTeX$$aARTICLE
001048850 3367_ $$2DataCite$$aOutput Types/Working Paper
001048850 520__ $$aThe equivariant behaviour of features is essential in many computer vision tasks, yet popular self-supervised learning (SSL) methods tend to constrain equivariance by design. We propose a self-supervised learning approach where the system learns transformations independently by reconstructing images that have undergone previously unseen transformations. Specifically, the model is tasked to reconstruct intermediate transformed images, e.g. translated or rotated images, without prior knowledge of these transformations. This auxiliary task encourages the model to develop equivariance-coherent features without relying on predefined transformation rules. To this end, we apply transformations to the input image, generating an image pair, and then split the extracted features into two sets per image. One set is used with a usual SSL loss encouraging invariance, the other with our loss based on the auxiliary task to reconstruct the intermediate transformed images. Our loss and the SSL loss are linearly combined with weighted terms. Evaluating on synthetic tasks with natural images, our proposed method strongly outperforms all competitors, regardless of whether they are designed to learn equivariance. Furthermore, when trained alongside augmentation-based methods as the invariance tasks, such as iBOT or DINOv2, we successfully learn a balanced combination of invariant and equivariant features. Our approach performs strong on a rich set of realistic computer vision downstream tasks, almost always improving over all baselines.
001048850 536__ $$0G:(DE-HGF)POF4-5111$$a5111 - Domain-Specific Simulation & Data Life Cycle Labs (SDLs) and Research Groups (POF4-511)$$cPOF4-511$$fPOF IV$$x0
001048850 536__ $$0G:(DE-HGF)POF4-5112$$a5112 - Cross-Domain Algorithms, Tools, Methods Labs (ATMLs) and Research Groups (POF4-511)$$cPOF4-511$$fPOF IV$$x1
001048850 536__ $$0G:(DE-Juel1)Helmholtz-SLNS$$aSLNS - SimLab Neuroscience (Helmholtz-SLNS)$$cHelmholtz-SLNS$$x2
001048850 7001_ $$0P:(DE-Juel1)128666$$aBruns, Benjamin$$b1$$ufzj
001048850 7001_ $$0P:(DE-Juel1)129394$$aScharr, Hanno$$b2$$ufzj
001048850 7001_ $$0P:(DE-Juel1)129347$$aKrajsek, Kai$$b3$$ufzj
001048850 8564_ $$uhttps://arxiv.org/abs/2503.18753
001048850 8564_ $$uhttps://juser.fz-juelich.de/record/1048850/files/2503.18753v1.pdf$$yOpenAccess
001048850 909CO $$ooai:juser.fz-juelich.de:1048850$$popenaire$$popen_access$$pVDB$$pdriver$$pdnbdelivery
001048850 9101_ $$0I:(DE-588b)5008462-8$$6P:(DE-Juel1)190396$$aForschungszentrum Jülich$$b0$$kFZJ
001048850 9101_ $$0I:(DE-588b)5008462-8$$6P:(DE-Juel1)128666$$aForschungszentrum Jülich$$b1$$kFZJ
001048850 9101_ $$0I:(DE-588b)5008462-8$$6P:(DE-Juel1)129394$$aForschungszentrum Jülich$$b2$$kFZJ
001048850 9101_ $$0I:(DE-588b)5008462-8$$6P:(DE-Juel1)129347$$aForschungszentrum Jülich$$b3$$kFZJ
001048850 9131_ $$0G:(DE-HGF)POF4-511$$1G:(DE-HGF)POF4-510$$2G:(DE-HGF)POF4-500$$3G:(DE-HGF)POF4$$4G:(DE-HGF)POF$$9G:(DE-HGF)POF4-5111$$aDE-HGF$$bKey Technologies$$lEngineering Digital Futures – Supercomputing, Data Management and Information Security for Knowledge and Action$$vEnabling Computational- & Data-Intensive Science and Engineering$$x0
001048850 9131_ $$0G:(DE-HGF)POF4-511$$1G:(DE-HGF)POF4-510$$2G:(DE-HGF)POF4-500$$3G:(DE-HGF)POF4$$4G:(DE-HGF)POF$$9G:(DE-HGF)POF4-5112$$aDE-HGF$$bKey Technologies$$lEngineering Digital Futures – Supercomputing, Data Management and Information Security for Knowledge and Action$$vEnabling Computational- & Data-Intensive Science and Engineering$$x1
001048850 9141_ $$y2025
001048850 915__ $$0StatID:(DE-HGF)0510$$2StatID$$aOpenAccess
001048850 920__ $$lyes
001048850 9201_ $$0I:(DE-Juel1)IAS-8-20210421$$kIAS-8$$lDatenanalyse und Maschinenlernen$$x0
001048850 9201_ $$0I:(DE-Juel1)JSC-20090406$$kJSC$$lJülich Supercomputing Center$$x1
001048850 9801_ $$aFullTexts
001048850 980__ $$apreprint
001048850 980__ $$aVDB
001048850 980__ $$aUNRESTRICTED
001048850 980__ $$aI:(DE-Juel1)IAS-8-20210421
001048850 980__ $$aI:(DE-Juel1)JSC-20090406