Technology Category
Computer Vision: SegmentationMachine Learning: Unsupervised & Self-Supervised LearningSearch and Optimization: Learning to Search
Leveraging Self-Supervised Features for Efficient Flooded Region Identification in UAV Aerial Images
🤖 AI Summary
To address the high annotation cost and poor generalization in semantic segmentation of flood regions from UAV-captured imagery, this paper proposes an encoder–decoder framework that integrates DINOv2 self-supervised visual features. We innovatively design two DINOv2 feature embedding architectures to effectively transfer generic representations—pretrained on non-aerial scenes—to the aerial flood detection task, substantially reducing reliance on pixel-level annotations. Under limited labeling conditions, our model achieves state-of-the-art segmentation accuracy (3.2–5.8% mIoU improvement) across multiple real-world UAV flood datasets, while maintaining strong generalization and deployment efficiency. This work empirically validates the transferability of general-purpose self-supervised vision models to few-shot remote sensing segmentation tasks, offering a scalable technical pathway for rapid post-disaster assessment.
Technology Category
Application Category
📝 Abstract
Identifying regions affected by disasters is a vital step in effectively managing and planning relief and rescue efforts. Unlike the traditional approaches of manually assessing post-disaster damage, analyzing images of Unmanned Aerial Vehicles (UAVs) offers an objective and reliable way to assess the damage. In the past, segmentation techniques have been adopted to identify post-flood damage in UAV aerial images. However, most of these supervised learning approaches rely on manually annotated datasets. Indeed, annotating images is a time-consuming and error-prone task that requires domain expertise. This work focuses on leveraging self-supervised features to accurately identify flooded regions in UAV aerial images. This work proposes two encoder-decoder-based segmentation approaches, which integrate the visual features learned from DINOv2 with the traditional encoder backbone. This study investigates the generalization of self-supervised features for UAV aerial images. Specifically, we evaluate the effectiveness of features from the DINOv2 model, trained on non-aerial images, for segmenting aerial images, noting the distinct perspectives between the two image types. Our results demonstrate that DINOv2's self-supervised pretraining on natural images generates transferable, general-purpose visual features that streamline the development of aerial segmentation workflows. By leveraging these features as a foundation, we significantly reduce reliance on labor-intensive manual annotation processes, enabling high-accuracy segmentation with limited labeled aerial data.
Problem
Research questions and friction points this paper is trying to address.
Identifying flooded regions in UAV aerial images efficiently
Reducing reliance on manual annotation for disaster assessment
Leveraging self-supervised features for accurate aerial image segmentation
Innovation
Methods, ideas, or system contributions that make the work stand out.
Self-supervised DINOv2 features for aerial segmentation
Encoder-decoder models with DINOv2 integration
Transfer learning from natural to aerial images
🔎 Similar Papers
No similar papers found.
