AI4SeaIce: selecting loss functions for automated SAR sea ice concentration charting

Kucik, Andrzej S.; Stokholm, Andreas

doi:10.1038/s41598-023-32467-x

Cited by 14 publications

(7 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In this study, we refer to several conclusions in their experiments in (Kucik and Stokholm, 2023;Stokholm et al, 2022;Stokholm et al, 2023), including that 1) NERSC noise correction is superior for U-Net model predicting full sea ice covers with little variation in the SAR textures, 2) 11 classes of SIC shows the highest detection accuracies of water and 100% SIC,…”

mentioning

confidence: 76%

“…More details can be found in the user manuals for the two datasets. Multi-series installments of articles related to AI4Arctic include (Esbensen, 2022;Stokholm et al, 2022;Stokholm et al, 2023;Kucik and Stokholm, 2023). In this paper, we used both the raw and ready-to-train datasets of the latest versions.…”

Section: Ai4arctic Project Datasetmentioning

confidence: 99%

“…Our U-Net experiments demonstrate successful ice-water classification, whereas achieving accurate multi-category SIC prediction remains a challenge. Kucik and Stokholm (2023) combined the ice discriminator (IceDisc) and intermediate discriminator (InetDisc) to improve the performance for intermediate classes however under the precision limitation of human labels. Improved SIC predictions are expected by using the SAR-Chart or SAR-AMSR2 fused SIC labels.…”

Section: Future Workmentioning

confidence: 99%

“…However, certain surfaces may not adhere strictly to these patterns, necessitating further experimentation for effective utilization of dual-polarization data. At least, it has been observed that dual-polarized data are sufficient to achieve high-precision icewater recognition using machine learning algorithms (Kucik and Stokholm, 2023;Wang and Li, 2020).…”

mentioning

confidence: 99%

See 3 more Smart Citations

Combining the U-Net model and a Multi-textRG algorithm for fine SAR ice-water classification

Sun,

Wang,

Cheng

et al. 2024

Preprint

View full text Add to dashboard Cite

Abstract. Sea ice classification faces challenges due to the similarity among surfaces such as wind-driven open water (OW), smooth thin ice, level first-year ice (FYI), and melted ice surfaces. Previous algorithms combine unsupervised region segmentation and supervised neural networks, yet struggle due to limited manual labels and inaccurate region segmentation. We propose to adopt a supervised neural network followed by a region segmentation algorithm with experiential knowledge involved to solve the ambiguous recognition question and sample number limitation. Provided by the AI4Arctic competition, the preprocessed GCOM-W1 AMSR2 36.5GHz H polarization and Sentinel-1 SAR EW dual-polarization data, the CIS/DMI ice chart labels, and the pre-trained U-Net CNN model are employed to perform semantic segmentation of ice and water with near-100 % accuracy. Subsequently, within the U-Net semantically segmented ice mask, a multistage pixel-based ice detection algorithm developed on GLCM textures of SAR images and region growing approach, the Multi-textRG algorithm, refines the ice edge details. We validate the results on Landsat-8 and Sentinel-2 optical data yielding an overall accuracy of 83.11 %, low false negative (FN) of 4.03 % indicating underestimated low backscatter ice surfaces and higher false positive (FP) of 12.86 % reflecting their resolution difference along edges. More importantly, we fused the SAR-based ice detection with CIS/DMI ice charts and AMSR2 ASI SIC product obtaining SAR-Chart and SAR-AMSR2 labels, which enhance ice edge depictions and SIC variation contours. Repeating the two-step procedure with the high-precision SIC labels demonstrates the U-Net model's capability to extract detailed ice edges information and stability of the Multi-textRG algorithm. The U-Net model trained on SAR-AMSR2 label achieves the highest R2-score of 91.993 %, the largest OWrecall (recall of OW) of 99.268 %, and large ov40recall (recall of ice with over-40 % SIC) of 99.207 %. Our algorithm framework solves the accurate ice-water classification at all seasons and facilitates the sample labelling for improving SIC estimation accuracy based on CNN models.

show abstract

mentioning

confidence: 76%

Section: Ai4arctic Project Datasetmentioning

confidence: 99%

Section: Future Workmentioning

confidence: 99%

mentioning

confidence: 99%

See 2 more Smart Citations

Combining the U-Net model and a Multi-textRG algorithm for fine SAR ice-water classification

Sun,

Wang,

Cheng

et al. 2024

Preprint

View full text Add to dashboard Cite

show abstract

“…Stokholm et al [19] used the AI4Arctic/ASIP v2 (ASID-v2, [11]) benchmark dataset to estimate sea ice concentration developing new variations of U-Net [20], a semantic segmentation architecture. Kucik and Stokholm [21] showed that the choice of loss function can significantly affect the appearance of the sea ice concentration predictions generated by CNN models.…”

Section: Related Workmentioning

confidence: 99%

Model Ensemble With Dropout for Uncertainty Estimation in Sea Ice Segmentation Using Sentinel-1 SAR

Pires de Lima,

Karimzadeh

2023

IEEE Trans. Geosci. Remote Sensing

View full text Add to dashboard Cite

Despite the growing use of deep learning in sea ice mapping with SAR imagery, the study of model uncertainty and segmentation results remains limited. Deep learning models often produce overconfident predictions, a concern in sea ice mapping where misclassification can impact marine navigation safety. We incorporate and compare dropout and model ensemble within a convolutional neural network segmentation architecture to highlight regions with prediction uncertainty, and explore the impact of loss function choice. We evaluate model generalization and uncertainty characterization by training and evaluating models on the AI4Arctic Sea Ice Challenge Dataset (primary). We further explore model uncertainty by testing the trained models on the Extreme Earth version 2 Dataset (secondary). The primary and secondary datasets vary in number of scenes as well as in the available data and preprocessing. We obtain test F1 results higher than 0.97 for the primary dataset. Although the F1 performance for the secondary dataset is reduced to 0.93, the generated sea ice maps are reasonable across several Sentinel-1 scenes, and our proposed strategy helps in identification of misclassified and uncertain regions for human quality control. Our models seem to be robust against banding noise in Sentinel-1 SAR, and the prediction uncertainty frequently highlights ice regions misclassified as water, indicating its potential for realworld applications. Our study advances the field of machine learning-based sea ice mapping and highlights the importance of uncertainty estimation and cross-dataset evaluation for model development and deployment. Our approach can be adopted for other remote sensing applications as well.

show abstract

Comparison of Cross-Entropy, Dice, and Focal Loss for Sea Ice Type Segmentation

de Lima,

Vahedi,

Karimzadeh

2023

IGARSS 2023 - 2023 IEEE International Geoscience and Remote Sensing Symposium

View full text Add to dashboard Cite

AI4SeaIce: selecting loss functions for automated SAR sea ice concentration charting

Cited by 14 publications

References 24 publications

Combining the U-Net model and a Multi-textRG algorithm for fine SAR ice-water classification

Combining the U-Net model and a Multi-textRG algorithm for fine SAR ice-water classification

Model Ensemble With Dropout for Uncertainty Estimation in Sea Ice Segmentation Using Sentinel-1 SAR

Comparison of Cross-Entropy, Dice, and Focal Loss for Sea Ice Type Segmentation

Contact Info

Product

Resources

About