MANGO: Learning Disentangled Image Transformation Manifolds with Grouped Operators
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2024
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866917991340310528 |
|---|---|
| author | Ancelin, Brighton Chen, Yenho Guan, Peimeng Kaushik, Chiraag Martin-Urcelay, Belen Saad-Falcon, Alex Singh, Nakul |
| author_facet | Ancelin, Brighton Chen, Yenho Guan, Peimeng Kaushik, Chiraag Martin-Urcelay, Belen Saad-Falcon, Alex Singh, Nakul |
| contents | Learning semantically meaningful image transformations (i.e. rotation, thickness, blur) directly from examples can be a challenging task. Recently, the Manifold Autoencoder (MAE) proposed using a set of Lie group operators to learn image transformations directly from examples. However, this approach has limitations, as the learned operators are not guaranteed to be disentangled and the training routine is prohibitively expensive when scaling up the model. To address these limitations, we propose MANGO (transformation Manifolds with Grouped Operators) for learning disentangled operators that describe image transformations in distinct latent subspaces. Moreover, our approach allows practitioners the ability to define which transformations they aim to model, thus improving the semantic meaning of the learned operators. Through our experiments, we demonstrate that MANGO enables composition of image transformations and introduces a one-phase training routine that leads to a 100x speedup over prior works. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2409_09542 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | MANGO: Learning Disentangled Image Transformation Manifolds with Grouped Operators Ancelin, Brighton Chen, Yenho Guan, Peimeng Kaushik, Chiraag Martin-Urcelay, Belen Saad-Falcon, Alex Singh, Nakul Image and Video Processing Computer Vision and Pattern Recognition Machine Learning I.2.6; I.4.2; I.4.7; I.4.10; I.5.1 Learning semantically meaningful image transformations (i.e. rotation, thickness, blur) directly from examples can be a challenging task. Recently, the Manifold Autoencoder (MAE) proposed using a set of Lie group operators to learn image transformations directly from examples. However, this approach has limitations, as the learned operators are not guaranteed to be disentangled and the training routine is prohibitively expensive when scaling up the model. To address these limitations, we propose MANGO (transformation Manifolds with Grouped Operators) for learning disentangled operators that describe image transformations in distinct latent subspaces. Moreover, our approach allows practitioners the ability to define which transformations they aim to model, thus improving the semantic meaning of the learned operators. Through our experiments, we demonstrate that MANGO enables composition of image transformations and introduces a one-phase training routine that leads to a 100x speedup over prior works. |
| title | MANGO: Learning Disentangled Image Transformation Manifolds with Grouped Operators |
| topic | Image and Video Processing Computer Vision and Pattern Recognition Machine Learning I.2.6; I.4.2; I.4.7; I.4.10; I.5.1 |
| url | https://arxiv.org/abs/2409.09542 |