ShapeEmbed: a self-supervised learning framework for 2D contour quantification
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2025
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866915368028602368 |
|---|---|
| author | Romero, Anna Foix Russell, Craig Krull, Alexander Uhlmann, Virginie |
| author_facet | Romero, Anna Foix Russell, Craig Krull, Alexander Uhlmann, Virginie |
| contents | The shape of objects is an important source of visual information in a wide range of applications. One of the core challenges of shape quantification is to ensure that the extracted measurements remain invariant to transformations that preserve an object's intrinsic geometry, such as changing its size, orientation, and position in the image. In this work, we introduce ShapeEmbed, a self-supervised representation learning framework designed to encode the contour of objects in 2D images, represented as a Euclidean distance matrix, into a shape descriptor that is invariant to translation, scaling, rotation, reflection, and point indexing. Our approach overcomes the limitations of traditional shape descriptors while improving upon existing state-of-the-art autoencoder-based approaches. We demonstrate that the descriptors learned by our framework outperform their competitors in shape classification tasks on natural and biological images. We envision our approach to be of particular relevance to biological imaging applications. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2507_01009 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | ShapeEmbed: a self-supervised learning framework for 2D contour quantification Romero, Anna Foix Russell, Craig Krull, Alexander Uhlmann, Virginie Computer Vision and Pattern Recognition Quantitative Methods The shape of objects is an important source of visual information in a wide range of applications. One of the core challenges of shape quantification is to ensure that the extracted measurements remain invariant to transformations that preserve an object's intrinsic geometry, such as changing its size, orientation, and position in the image. In this work, we introduce ShapeEmbed, a self-supervised representation learning framework designed to encode the contour of objects in 2D images, represented as a Euclidean distance matrix, into a shape descriptor that is invariant to translation, scaling, rotation, reflection, and point indexing. Our approach overcomes the limitations of traditional shape descriptors while improving upon existing state-of-the-art autoencoder-based approaches. We demonstrate that the descriptors learned by our framework outperform their competitors in shape classification tasks on natural and biological images. We envision our approach to be of particular relevance to biological imaging applications. |
| title | ShapeEmbed: a self-supervised learning framework for 2D contour quantification |
| topic | Computer Vision and Pattern Recognition Quantitative Methods |
| url | https://arxiv.org/abs/2507.01009 |