Unsupervised CD in Satellite Image Time Series by Contrastive Learning and Feature Tracking

Unsupervised change detection using contrastive learning has significantly improved the performance of literature techniques. However, at present it only focuses on the bi-temporal change detection scenario. Previous state-of-the-art models for image time-series change detection have traditionally depended on features obtained either through clustering learning or by training models from scratch using pseudo labels tailored to each scene. However, these approaches fail to either exploit the spatial-temporal information of image time-series or generalize to unseen scenarios. In this work, we propose a two-stage approach to unsupervised change detection in satellite image time-series using contrastive learning with feature tracking. By deriving pseudo labels from pre-trained models and using feature tracking to propagate them within the image time-series, we improve the consistency of our pseudo labels and address the challenges of seasonal changes in long-term remote sensing image time-series. We adopt the self-training algorithm with ConvLSTM on the obtained pseudo labels, where we first use supervised contrastive loss and contrastive random walks to further improve the feature correspondence in space-time. Then a fully connected layer is fine-tuned on the pre-trained multi-temporal features for generating the final change maps. Through comprehensive experiments on two datasets, we demonstrate consistent improvements in accuracy on fitting and inference scenarios.