🤖 AI Summary
This study addresses the automatic morphological classification of eclipsing binary light curves. We propose an image-based representation method that combines polar-coordinate mapping with hexagonal binning (hexbin) visualization to transform one-dimensional phase-folded light curves into two-dimensional images exhibiting structural robustness and scale invariance. Building upon this representation, we design a two-stage hierarchical classification framework: the first stage distinguishes detached from overcontact systems; the second stage identifies the presence of starspots. We employ transfer learning using pretrained ResNet50 and ViT-Base (patch16_224) models, fine-tuned on synthetic data and applied to multi-band observations. Experiments on real-world datasets—OGLE, DEBCat, and WUMaCat—achieve overall classification accuracies of 94%–100%, substantially outperforming conventional time-series feature-based approaches. Binary configuration identification accuracy exceeds 96%, whereas starspot detection remains challenging.
📝 Abstract
We present an application of computer vision methods to classify the light curves of eclipsing binaries (EB). We have used pre-trained models based on convolutional neural networks ($ extit{ResNet50}$) and vision transformers ($ extit{vit_base_patch16_224}$), which were fine-tuned on images created from synthetic datasets. To improve model generalisation and reduce overfitting, we developed a novel image representation by transforming phase-folded light curves into polar coordinates combined with hexbin visualisation. Our hierarchical approach in the first stage classifies systems into detached and overcontact types, and in the second stage identifies the presence or absence of spots. The binary classification models achieved high accuracy ($>96%$) on validation data across multiple passbands (Gaia~$G$, $I$, and $TESS$) and demonstrated strong performance ($>94%$, up to $100%$ for $TESS$) when tested on extensive observational data from the OGLE, DEBCat, and WUMaCat catalogues. While the primary binary classification was highly successful, the secondary task of automated spot detection performed poorly, revealing a significant limitation of our models for identifying subtle photometric features. This study highlights the potential of computer vision for EB morphological classification in large-scale surveys, but underscores the need for further research into robust, automated spot detection.