A Data-Centric Vision Transformer Baseline for SAR Sea Ice Classification

• URL: http://arxiv.org/abs/2604.03094v1
• Date: Fri, 03 Apr 2026 15:18:00 GMT
• Title: A Data-Centric Vision Transformer Baseline for SAR Sea Ice Classification
• Abstract summary: We use the AI4Arctic/ASIP Sea Ice dataset (v2), which contains 461 Sentinel-1 scenes matched with expert ice charts.<n>We compare ViT-Base models trained with cross entropy and weighted cross-entropy against a ViT-Large model trained with focal loss.<n>These results show that focal-loss training offers a more useful precision-recall trade-off than weighted cross-entropy for rare ice classes.
• Score: 2.4851820343103035
• License: http://creativecommons.org/licenses/by-nc-sa/4.0/
• Abstract: Accurate and automated sea ice classification is important for climate monitoring and maritime safety in the Arctic. While Synthetic Aperture Radar (SAR) is the operational standard because of its all-weather capability, it remains challenging to distinguish morphologically similar ice classes under severe class imbalance. Rather than claiming a fully validated multimodal system, this paper establishes a trustworthy SAR only baseline that future fusion work can build upon. Using the AI4Arctic/ASIP Sea Ice Dataset (v2), which contains 461 Sentinel-1 scenes matched with expert ice charts, we combine full-resolution Sentinel-1 Extra Wide inputs, leakage-aware stratified patch splitting, SIGRID-3 stage-of-development labels, and training-set normalization to evaluate Vision Transformer baselines. We compare ViT-Base models trained with cross entropy and weighted cross-entropy against a ViT-Large model trained with focal loss. Among the tested configurations, ViT-Large with focal loss achieves 69.6% held-out accuracy, 68.8% weighted F1, and 83.9% precision on the minority Multi-Year Ice class. These results show that focal-loss training offers a more useful precision-recall trade-off than weighted cross-entropy for rare ice classes and establishes a cleaner baseline for future multimodal fusion with optical, thermal, or meteorological data.

Related papers

• V-JEPA 2.1: Unlocking Dense Features in Video Self-Supervised Learning [46.89986408533846]
  V-JEPA 2.1 is a family of self-supervised models that learn dense, high-quality visual representations for both images and videos.<n>We show that V-JEPA 2.1 significantly advances the state-of-the-art in dense visual understanding and world modeling.
  arXiv  Detail & Related papers  (2026-03-15T17:02:40Z)
• Breaking the Statistical Similarity Trap in Extreme Convection Detection [0.0]
  Current evaluation metrics for deep learning weather models reward blurry predictions while missing rare, high-impact events.<n>We introduce DART, a framework addressing the challenge of transforming coarse atmospheric forecasts into high-resolution satellite temperature fields.<n>DART employs dual-decoder architecture with explicit background/extreme decomposition, physically motivated oversampling, and task-specific loss functions.
  arXiv  Detail & Related papers  (2025-09-11T07:10:45Z)
• Machine Learning-Based Classification of Vessel Types in Straits Using AIS Tracks [0.0]
  This paper presents a machine-learning pipeline that classifies moving vessels using only AIS data.<n>We analyze eight days of historical AIS from the Danish Maritime Authority covering the Bornholm Strait in the Baltic Sea.
  arXiv  Detail & Related papers  (2025-09-09T19:52:43Z)
• ReconMOST: Multi-Layer Sea Temperature Reconstruction with Observations-Guided Diffusion [48.540756751934836]
  ReconMOST is a data-driven guided diffusion model framework for multi-layer sea temperature reconstruction.<n>Our method extends ML-based SST reconstruction to a global, multi-layer setting, handling over 92.5% missing data.
  arXiv  Detail & Related papers  (2025-06-12T06:27:22Z)
• Pseudo-Label Guided Real-World Image De-weathering: A Learning Framework with Imperfect Supervision [57.5699142476311]
  We propose a unified solution for real-world image de-weathering with non-ideal supervision.<n>Our method exhibits significant advantages when trained on imperfectly aligned de-weathering datasets.
  arXiv  Detail & Related papers  (2025-04-14T07:24:03Z)
• Efficient Self-Supervised Learning for Earth Observation via Dynamic Dataset Curation [67.23953699167274]
  Self-supervised learning (SSL) has enabled the development of vision foundation models for Earth Observation (EO)<n>In EO, this challenge is amplified by the redundancy and heavy-tailed distributions common in satellite imagery.<n>We propose a dynamic dataset pruning strategy designed to improve SSL pre-training by maximizing dataset diversity and balance.
  arXiv  Detail & Related papers  (2025-04-09T15:13:26Z)
• Assessing Foundation Models for Sea Ice Type Segmentation in Sentinel-1 SAR Imagery [2.255599116578844]
  This study evaluates ten foundation models (FMs) for sea ice type segmentation using Sentinel-1 imagery.<n>Among the selected models, Prithvi-600M outperforms the baseline models, while CROMA achieves a very similar performance in F1-score.
  arXiv  Detail & Related papers  (2025-03-28T15:21:08Z)
• Comparison Study: Glacier Calving Front Delineation in Synthetic Aperture Radar Images With Deep Learning [11.298727820998566]
  This study presents the first comparison of Deep Learning systems on a common calving front benchmark dataset.<n>The best DL model's outputs deviate 221 m on average, while the average deviation of the human annotators is 38 m.<n>This significant difference shows that current DL systems do not yet match human performance and that further research is needed to enable fully automated monitoring of glacier calving fronts.
  arXiv  Detail & Related papers  (2025-01-09T14:43:36Z)
• Partial Label Learning with Focal Loss for Sea Ice Classification Based on Ice Charts [2.0270474485799017]
  We present a novel GeoAI approach to training sea ice classification by formalizing it as a partial label learning task with explicit confidence scores. We treat the polygon-level labels as candidate partial labels, assign the corresponding ice concentrations as confidence scores to each label, and integrate them with focal loss to train a Convolutional Neural Network (CNN) Our proposed approach leads to enhanced performance for sea ice classification in Sentinel-1 dual-polarized SAR images, improving classification accuracy (from 87% to 92%) and weighted average F-1 score (from 90% to 93%) compared to the conventional training approach.
  arXiv  Detail & Related papers  (2024-06-05T22:49:30Z)
• An Experimental Study on Exploring Strong Lightweight Vision Transformers via Masked Image Modeling Pre-Training [51.622652121580394]
  Masked image modeling (MIM) pre-training for large-scale vision transformers (ViTs) has enabled promising downstream performance on top of the learned self-supervised ViT features. In this paper, we question if the textitextremely simple lightweight ViTs' fine-tuning performance can also benefit from this pre-training paradigm. Our pre-training with distillation on pure lightweight ViTs with vanilla/hierarchical design ($5.7M$/$6.5M$) can achieve $79.4%$/$78.9%$ top-1 accuracy on ImageNet-1
  arXiv  Detail & Related papers  (2024-04-18T14:14:44Z)
• Automating global landslide detection with heterogeneous ensemble deep-learning classification [44.99833362998488]
  Landslides threaten infrastructure, including roads, railways, buildings, and human life. Hazard-based spatial planning and early warning systems are cost-effective strategies to reduce the risk to society from landslides. Deep learning models have recently been applied for landside mapping using medium- to high-resolution satellite images as input.
  arXiv  Detail & Related papers  (2023-09-12T10:56:16Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.