What question did this study set out to answer?

To develop an edge-guided and uncertainty-aware hybrid CNN-Transformer framework that accurately segments marine oil spills in synthetic aperture radar images despite speckle noise and look-alike features.

August 19, 2026Discover ComputingOpen Access

An edge-guided hybrid CNN–transformer framework for accurate marine oil spill segmentation in synthetic aperture radar images

View Full Paper

Ask AI

Bookmark

Authors

JJJeena Joseph

Discussion

Member takes

Overview

Deep learning evaluation demonstrates superior oil spill segmentation in synthetic aperture radar imagery, indicating enhanced boundary detection and fewer false alarms.

Key Points

To develop an edge-guided and uncertainty-aware hybrid CNN-Transformer framework that accurately segments marine oil spills in synthetic aperture radar images despite speckle noise and look-alike features.
Designed a hybrid architecture combining a convolutional neural network encoder for local texture capture, a transformer for long-range context, and a cross-attention fusion mechanism.
Integrated an edge guidance module to reinforce spill boundaries and an uncertainty refinement head to resolve ambiguous image regions.
Trained and evaluated the model on a synthetic aperture radar oil spill dataset containing 6,455 training images and 1,615 testing images.
The proposed framework achieved a Dice score of 0.942, IoU of 0.891, precision of 0.951, recall of 0.928, and accuracy of 0.968.
The model outperformed baseline architectures including U-Net, DeepLabV3+, Attention U-Net, and TransUNet while demonstrating sharper boundary preservation and reduced false detections.

Cite This Study

Jeena Joseph (2026) studied this question.

synapsesocial.com/papers/6a8563c403308d306e2d7139 https://doi.org/10.1007/s10791-026-10418-0

View Full Paper

Ask AI

Bookmark

August 19, 2026Discover ComputingOpen Access

An edge-guided hybrid CNN–transformer framework for accurate marine oil spill segmentation in synthetic aperture radar images

View Full Paper

Ask AI

Bookmark

Authors

JJJeena Joseph

Discussion

Member takes

Overview

Deep learning evaluation demonstrates superior oil spill segmentation in synthetic aperture radar imagery, indicating enhanced boundary detection and fewer false alarms.

Key Points

To develop an edge-guided and uncertainty-aware hybrid CNN-Transformer framework that accurately segments marine oil spills in synthetic aperture radar images despite speckle noise and look-alike features.
Designed a hybrid architecture combining a convolutional neural network encoder for local texture capture, a transformer for long-range context, and a cross-attention fusion mechanism.
Integrated an edge guidance module to reinforce spill boundaries and an uncertainty refinement head to resolve ambiguous image regions.
Trained and evaluated the model on a synthetic aperture radar oil spill dataset containing 6,455 training images and 1,615 testing images.
The proposed framework achieved a Dice score of 0.942, IoU of 0.891, precision of 0.951, recall of 0.928, and accuracy of 0.968.
The model outperformed baseline architectures including U-Net, DeepLabV3+, Attention U-Net, and TransUNet while demonstrating sharper boundary preservation and reduced false detections.

Cite This Study

Jeena Joseph (2026) studied this question.

synapsesocial.com/papers/6a8563c403308d306e2d7139 https://doi.org/10.1007/s10791-026-10418-0

Also Consider

Synapse has enriched 4 closely related papers on similar clinical questions. Consider them for comparative context:

View Full Paper

Ask AI

Bookmark